Interestingly it was very clear at around $10MM ARR that DO was on a trajectory to IPO. You can thank Moisey Uretsky for a fantastic product idea and his brother Ben for CEOing it for so long. Congratulations to everyone who was involved in building DigitalOcean, it was an INCREDIBLY wild ride in the early days, lots of chaos but through all the chaos and disfunction, I think everyone involved knew this day would come.
Personally I'm really proud of the work we did, and I'm overjoyed to have been part of building a business that is going to be a public company. Thanks to the HN community for being so supportive over the years.
The thesis of DigitalOcean was literally “this very established market needs a venture capital growth play because we saw Slicehost exit to Rackspace”. In the beginning the entirety of DO’s offering, tech for tech, was almost a direct lift of Linode and, as they scaled, made all the same technical mistakes and learning experiences that Linode did.
I guess you could call that a fantastic product idea. I’d call it applying an aggressive exit strategy to Chris Aker’s fantastic product idea from 2003 and further commoditizing said product in its wake.
Actually, Ben and Moisey owned a managed hosting company that was coming under a lot of pressure from rackspace. DigitalOcean as a product was a direct response to that market pressure. I think if you ask any of the early executives, none of us loooovveed the venture component, it was a way to grow the business that was already growing rapidly with strong product market fit. Remember, a16z only invested once we were doing millions and millions in ARR. In terms of Linode, I have no idea what you're talking about. DigitalOcean had 1 click $5 SSD offerings years before linode/anyone else made any changes to their business.
I remember Linode being hesitant to dip below $20 because the thinking was that it would diminish the quality of support. That thinking was reinforced when you started with $10 (if I recall, it’s been nearly a decade) and we started getting large numbers of refugees burned by experiences with your support organization. Neighboring comments tell me it’s still a problem. We consciously didn’t want to grow $5 fast because we didn’t have capital to throw at any problem that developed. We, the market, not just Linode, also knew competing to the bottom on price largely relegated shared/cpanel/etc hosting to $0.99 commodity status and we weren’t keen to kick off such a war for obvious reasons. (That was quite prescient, looking back from outside in 2021.)
Anyway, yes, you eventually reached a position where you dictated the pricing structure, slightly annoyingly because it was a product that looked remarkably similar to ours (down to the incentivized Linode Library clone that’s getting praise elsewhere in this thread). I wasn’t talking about changes you forced. The entire product top to bottom was referenced against Linode’s work to grow you to a position to force those changes. Sorry if I wasn’t clear. I’m also not going after you for it; I’m only responding to fantastic product idea.
It’s just odd that when competitors would ask me my thoughts on the “Linode clone” at 2011-era conferences that’s something that’s news to you. Is that really not apparent to an executive? If not that you did so, at least that seemingly the entire incumbency, not just Linode, thought so?
> it was a product that looked remarkably similar to ours
As a developer, DO never looked anything like it. Linode was expensive and on the same playfield as Rackspace and a ton of other cloud hosts, very different from the instantly-spin-up-a-tiny-vm-for-five-bucks model. Their web UI also has always been miles ahead.
The competition also didn’t have SSD as an option until much later, and that was a HUGE selling point for digital ocean at the time.
> As a developer, DO never looked anything like it.
The UI, the simplicity, SSDs, additional block storage... but also the API and its doc, as well as IPv6, regions, snapshots/backups, and availability. Later on, load balancers, object storage and k8s. All of that much simpler than AWS and both better thought out and more featureful than Linode or OVH.
The simper than AWS aspect was one of the main reasons I choose DO to do most of my cloud stuff. Also AWS had this odd habit of not auto charging me so it would go a couple months then I would get hit with multiple months worth of a bill.
I’m customer for years and I don’t use any of the features you mentioned past the first two; I’m part of the 5$ VM crowd for who this is the killer feature. Should OVH have provided on-demand instead of yearly plans for VPS I would have stayed there as I’ve my domains there already.
DO definitely had UI, SSD first. But Block Storage, Regions, Snapshots Backup and Load Balancer all were either first on Linode or being worked on for a long time. Linode also had Pooled Transfer and other tidbit for a long time well before DO.
Remember we are talking about early days of DO. Of course later DO innovate more and Linode had to follow. And the new Linode UI is arguably better than DO. Although it did took them years to make it. The beauty of competition.
Before I ever paid DO a cent, I made significant use of a lot of their incredible documentation on setting up various servers, setting up your confide correctly, etc.
Right! Linode was $20/month at least and more focused to businesses. Not the then unbelievable $5/month for SSH ssh accessible virtual box with root access that DO started. Plus the SSD. Other hosting providers later caught on with DO.
But the thing I liked best was their "no hidden cost". You pay what you agreed before to pay. No surprise. No upsell, trick sell which were common with all hosting at that time.
DO still maintains that honesty. Not sure what will happen after their IPO.
I'd strongly disagree (been on DO since about 2012 and on Linode a few years earlier; I still use both). DO UI is ok these days but years ago Linode UI was far better in every way.
I remember at that time feeling almost bad for Linode because they had built up so much valuable educational content about what you could run on their services, and I was using it step-by-step on DO. The only differentiator was the price. Of course DO quickly replicated and excelled in providing that tutorial value as well.
Negative. Slicehost launched in the middle of Linode pivoting virtualization stacks after proving out the business and running into limitations of their first stack (UML). Slicehost was ... 2006, if memory serves? Linode was a few years old by that point. For those keeping score, Linode then went to Xen and now KVM just like everyone else.
Linode and Slicehost were very similar stories, though. Slicehost bootstrapped as well and was run by good people.
Whether your claim is true or not, there’s more to a product than the underlying technology. Frankly, this line of discussion reads as sour grapes and reflects poorly on the Linode team.
> more to a product than the underlying technology
Back when I was just hacking about on some SaaS ideas I looked into both Linode and DO. The positioning of both seemed quite separate to me, and DO "felt" more like the product I should be using.
Whether or not the tech was different, or that feeling was justified by anything other than an initial veneer is well up for debate - but entirely agree, they're discreet products, and this doesn't really reflect well on Linode.
I don't think it reflects poorly on them at all. When DO started everyone still used the help document of Linode and Slicehost to setup and spin a server on DO. They could do that because DO was a clone of these services. To their credit DO improved and grew, and that's no mean feat.
Yes. For those of us that actually went through the Slicehost, Linode era, the time when paying per month and not per hour for a simple VPS before everything was rolled into the word Cloud. Those comment read more like an accurate account of history while others read it as an attack of a company.
I've been a customer of both for years -- in Linode's case, since 2009. Your comments are coming across like sour grapes, so I think this is a good opportunity to hear about some of Linode's mistakes from a longtime Linode customer.
First, you guys had the customer support interface compromise in 2013, and the worst part about that was the way that it was handled by your organization. Linode wouldn't own up to the compromise initially, dragged their feet on announcing anything about it, and then tried to downplay it.
Then there was another similar incident in 2016, and, amazingly, the response from Linode was nearly identical to 2013. No lessons learned. After that, I started seeing influential people on HN say things like, "don't use Linode". I stayed on Linode, but I also started using DigitalOcean after the 2013 compromise and split my hosted services evenly between the two of you after 2016.
DigitalOcean aggressively iterated multiple aspects of their products in that time period. They announced block storage in 2016. Your customers immediately started asking when Linode would come out with a competing service, and that didn't happen until 2018. Here, it was Linode rushing to build a DigitalOcean feature.
Linode initially focused on expanding RAM and disk space for your customers' instances at the same price points. Every year, you'd announce, "hey, we're doubling your RAM!" or some such thing. It was awesome, but I also thought at the time that you were shooting yourselves in the foot. How many customers looked at that and went, "yep, cool, now I can downsize my Linode and give them less money each month"? I know I did, at least a couple of times.
Then, you chose to try to compete with DigitalOcean on price. As a long time Linode customer, I never wanted that. If I want a $5 VPS, I'll just go to DigitalOcean. Hell, you shouldn't even want me as a customer for $5. Let them deal with me instead. I wanted Linode to be the larger, slightly more expensive provider with the even better support.
I've had to deal with Linode support a few times over the years. They've mostly been awesome and first class. I've never been unhappy enough with Linode to want to leave altogether. Both of you should be focused on eating AWS, not each other; showing up to a competitor's HN thread to complain about them being copycats is really poor form.
You might be experiencing a form of survivorship bias / confirmation bias here.
If you work at Linode, presumably people would come up to you asking about something they then couched to you as a Linode clone. That doesn't mean it's how people were talking about DO when not directly speaking with someone at Linode.
I don’t understand how that impacts my point at all. If an opinion is only held in one context it’s not actually held? Is that what you’re saying? I’m sorry, I don’t follow.
Before DO VPS was (and somewhat remains) a community even among competition. I’m not basing what I’m saying solely on the example given.
He is saying that your perspective as a Linode employee might not represent how others saw Digital Ocean at the time. Which for me is glaringly obvious from your comments.
Well a non-DO employee (and someone who ran things on linode and then on DO) I would say its a fair read and what the mindshare at the time was.
Linode had decent service but honestly the irc channel was pretty toxic to noobs (one of the reasons I left!) and finding a "cheaper linode option" was great.
That might have switched quickly, but it was definitely a thing.
I've noticed that the folks who aren't so attentive to product / market fit and other items (customer ergonomics maybe?) tend not to really see what they are missing, what's different about the "other" guy that means they are in some ways cleaning up. I wonder if that describes linode a bit.
Security:
One issue early on for Linode was just abysmal security despite claims. Lots of denial and lack of transparency too which was even worse. For a biz focused offering that was just a no go.
Even 8 year ago there was stuff with the bitcoin hacks? Do folks remember any of this?
They might have been using cold fusion or something because they got hacked again I think just like a year later?
These weren't just little corner case hacks / issues - but major with plenty of denial and obfuscation from linode.
I'm sure I'm missing others (unsecured mysql databases etc?).
---
Anyways, at least in my world linode was actually seen as the cheap/crappy offering even though they actually cost more which was weird. But I def told folks to stay away just because of the repeated lack of care around security.
I had no idea DO was at 350M+ ARR - nice job by them - wow!
I guess you may think of that in the same way the incumbents considered Amazon “another e-commerce play” or Google as “another search engine” while being blind to the real changes happening underneath.
Perhaps it wasn’t their own mental model of the landscape at all, but just what they considered to be the most expedient way of getting the message across in terms customers would understand.
Hypothetical analogy: if I were describing my very-first-ever PaaS service, to competitors who’d only ever heard of IaaS services — and this wasn’t the point of the conversation (i.e. I wasn’t trying to sell them on the benefits of PaaS), but rather just a supporting statement to talking about how we do/don’t offer an IaaS feature, because we’re a PaaS — then I’d probably describe my PaaS by analogy to some popular IaaS that has a feature-set closest to a PaaS.
In my own mental model, I have a separate node for “PaaS”; but if I know that the people I’m communicating with don’t — and teaching them what a PaaS is would take time away from the real topic we’re trying to focus on — then I’m going to describe my PaaS provider as “basically like $foo IaaS provider” when talking about how it has the same feature X. To them.
Again, I’m not basing what I’m saying solely on the conference example. I’m not arguing over the psychology of multiple conference conversations from ten years ago any further because it’s completely immaterial to my point (and one example of many; the point of the example was that it was openly discussed among competition, not “let’s go to the mat on what constitutes an actual opinion with the assumption that I don’t perceive conversation appropriately”).
Have you ever watched Kitchen Nightmares? Bad chef's are often surprised to get told their food tastes bad because everyone always tells them how good their food is!
If you know someone is from a particular company, you don't badmouth the company in front of them, but you might badmouth the competitors.
As a user, I agree - DO was seen a cheaper clone of Linode / Slicehost when they launched. I used to usually recommend Pair and Linode to my clients and, later DO to my more cost-conscious clients. At that time, I preferred Linode over DO, because the support was better and the prevalent practice than was to look suspiciously at new hosts because a lot of them folded up. You looked to older hosts, like Pair (and later Rackspace and Linode) because you knew they were reliable in both tech and the hosting business.
They still seem to be doing really well - last heard, they've migrated most of their servers from FreeBSD to Linux and also offer Linode and DO like VPS hosting.
As an original Slicehost customer, and long time Digital Ocean and Linode customer, your version of history is bullshit. I was looking to move from Rackspace after the Slicehost aquisition. Moved to Digital Ocean because of the SSD offering. It performed wonderfully. Needed bigger servers, and Linode had just created an offering competitive with Digital Ocean. I signed up and ran some benchmarks and Linode was about 25% faster than DO. I moved everything to Linode and the company has been there ever since. For what it is worth, my Linode bill is currently about $2000, and DO is $40. I am calling you out as a loyal Linode customer who thinks Linode is great.
I was a huge Linode fan for years. I used them exclusively. DO showed up and slowly but surely started being the better deal. Lower entry points, better specs, and then finally a more robust product line. I slowly found myself going to DO for new servers and eventually moved everything there for simplicities sake.
Linode still has a special place in my heart, but they have some work to do if they want to remain competitive.
Wow - really? I use both Linode and DO extensively and Linode is better on almost all fronts in my experience. Faster servers, better network, Linode's Object Storage is way better than DO's "spaces" and Linode's support is hands down the better of the two. If you're going to spend $5 on a server why would you go with DO unless you're already there?
> If you're going to spend $5 on a server why would you go with DO unless you're already there?
I would go with anyone other than Linode, considering they left the control plane running a publicly accessible unpatched version of ColdFusion, which led to multiple instances of them losing control of customer's data.
I mean that was terrible, but I know full well that every company ever will have dumb vulnerabilities occasionally, so by that metric I could never use anyone.
Last time I provision on linode and do at the same time a created a ubuntu box. I found I had to setup a few additional things in linode. Linode never felt as polished which is fine.
The one thing that made do better was they had a datacenter in my region.
No one mentioned vultr killer $2.50 which is where I would go if I wanted a dev box to play around on.
I remember something being off in Vultr $2.50 offer which made me stop halfway during sign-up. Had to go to https://www.vultr.com/products/cloud-compute/ again to remind myself: $2.50 VPS does not come with an IPv4 address. At least today they openly mark it and I think 5 years ago you'd only find it after a few clicks.
This was an advantage to me at the time. My ISP (Internode) had enabled dual-stack v6, and a few of the other options I was looking at on LowEndBox came with a NATed v4 address. Here was Vultr offering an IP I could fully control, even point real domains to!
I don't know if this is still the case, but I also got no SSH logspam from bots trying to log in on a v6 host.
Same here, I use both but prefer Linode in every aspect. Much better support and reliability in particular.
I moved a few very low-end hosts to DO when the $5 price point appeared but have never been overly impressed. Later Linode matched prices so no real reason to look into DO anymore.
But I do still use both. Partly to avoid migrating the DO hosts I have but more because I just like to keep aware of how both are evolving.
Linodes support is great; I can't say much about DO because I haven't had to use them.
Around the time I switched there was a lot going on. DO had released their $5 droplets. Linode was still stuck on their old, clunky UI, and the migration to the new "Cloud UI" wasn't executed well. Then there was a few DDOS attacks that took stuff down, and the multiple serious security vulnerabilities that they did a -horrible- job managing.
I've been at DO for a number of years now and it's been solid. I have a few dozen servers and take advantage of a few of their services. Their Managed Databases are incredibly well executed.
I've been happy over all so no reason to look elsewhere.
The real story is Ben and Moisey were running one of the largest hosting co's in porn and then there was a security problem via one of their upstream providers and it created a DISASTER. Thats how DO was born from what I understand.
As a user of both services for years, the only point on which they are similar was/is that they both offer shared cheap VMs (as do lots of other services).
Presenting early DO as a ripoff of linode is frankly absurd, the UI, user experience, API, Load balancing, backups, and docs were all vastly superior to Linode, and it honestly felt like the copying was more the other way round as Linode tried to catch up with their UI.
I'm puzzled by the praises on DO UI, since in the early days in particular I always found it intolerably terrible. Far less useful than Linode UI. Nowadays it's ok, but Linode is still has better UI.
I was on Linode for a while and moved to DigitalOcean. At the time it was a very simple calculus.
DigitalOcean was less established, less mature, offered fewer CPUs and less RAM per dollar and had only two (or was it three?) locations.
But they had SSDs and Linode did not.
The difference in Wordpress performance was night and day. So I switched and I'm still there. Linode dragged its feet on SSDs. I don't care why or how. I cared that I could get what I needed elsewhere.
What's the difference like today? As a long time Linode customer, I just never even felt the need to shop around. Maybe I should! It's always been just fine for my "tiny duty" Wordpress host. I always knew DO existed, but figured it's the same $5 and I'm probably getting about the same thing so why bother switching. I guess I'm a good example of why stickiness matters.
Today the pricing and services are essentially matched. Someone will obviously fight be on this statement, but I think today they are essentially equal from a cost perspective. They all run SSDs and have similar cost per RAM/CPU core.
But as others have said, Linode was around forever. Digital Ocean came in and disrupted them. When DO first came on the scene then you easily could have switched and got more for your money. But Linode has caught up and matched them. Right now they are both good. I have actually been favoring Linode lately with their new interface. But I have $3,000 or so of Digital Ocean credit that I am trying to use up first.
This doesn’t sound right. I know we evaluated linode vs do in 2016, and dropped linode because of lack of volume support. Digitalocean had that, even if only in FRA1, but it surely was ahead of linode back then
(Not sure if block storage came even earlier to other DO data centers, but as those were outside the EU they wouldn’t have been relevant)
I always wondered if they would eventually make a deliberate attempt to take over the "shared Cpanel hosting" market with a non-shared-host, non-cpanel model. It seems ripe for change...lots of little players.
The market was considerably too small and it would mean building to much towards the past. The thesis of "millions of *new* developers are coming online over the next 20 years" was the main north star.
It is small compared to cloud, but it's a ~$22 billion / year market. Compare to IaaS at ~$50b. So it's not tiny, but it is highly fragmented.
And I think the solution could be layered on top of what they already have. Cpanel is just a bunch of buttons, icons, and little apps to create web server instances and tie them to domains, database instances, backups, web file management, manage DNS records, and so on. Basically just a consistent web app to put on top of their "droplet" ecosystem. Much of it already exists, so a fair amount of the effort would just be marketing it.
Their approach to making VPS/Cloud offerings simple and approachable are really what set them apart, in my mind. I run all of my simple things there, because sometimes I don't want or need the complexity that actually managing things in Azure, GCloud, or AWS require.
I remember interviewing with DO back in late 2014 and it was nuts! The entire interview process took ~3 months and it was clear that they were defining the process as they went. The entire process was basically:
1. initial call with recruiter
2. homework project
3. call with engineer to discuss said homework project
4. two phone screens with engineers
5. onsite interview with 6 engineers
Finally I had was at the final step which was a call with Ben who was really interesting to talking to as we shared the sysadmin background.
They we going through some pretty crazy growth at the time so I forgave a lot of disorganization in the interview process. Unfortunately I didn't end up getting the role but I'm glad to have had that experience; it was definitely an interesting point in DigitalOcean's history.
> 2. homework project
> 3. call with engineer to discuss said homework project
How much time does it typically take (or should take) to complete a 'homework' project?
> 4. two phone screens with engineers
> 5. onsite interview with 6 engineers
This seems very 'camel is a horse by committee' to me.
Reminds me a bit (know it's different) of dating where someone wants you to meet their family for approval. I immediately pass on those dates (and will add I am happily married to a woman who did not do that).
I wonder about situations where it takes so many inputs to make a decision to me that speaks a great deal about the decision making process being faulty.
Fwiw when I graduated college years ago I was rejected by a company (friend of the family no less) who said they couldn't hire me because 'yes you are smart but you don't know this business' (true I knew zero). So I started a company like theirs myself and now many years later they are gone and my company (which I sold) still survives. (Will add this was years before the internet and common practice for people who knew nothing to go into businesses they knew about).
Now yes I do understand programming is not the same but my point still stands why so many chefs in the kitchen? Does anyone ever go back and track the people who were rejected what happened to them?
> How much time does it typically take (or should take) to complete a 'homework' project?
I was still in college at the time but I maybe spent 5-10 hours over the course of a week or so. I do remember it took a very long time for them to respond after I had sent it in though. I believe the project was to implement a basic web crawler in Go.
> Now yes I do understand programming is not the same but my point still stands why so many chefs in the kitchen?
After the homework problem I think that's where they were defining the process as they went. They seemed to be growing quite a bit at the time (IIRC when I was onsite they had just rented two more floors of office space) and I had bounced between a few contacts that were brand new to the company. The real struggle is that I never knew how many more steps were to come, and I'm not sure they knew either.
Though after the onsite they e-mailed me almost immediately to setup the call with Ben so I think everything had gone well up to that point. I don't think the call with Ben went poorly so my best guess is that they went with someone that had experience rather than a new grad.
FWIW I'm not actually bitter about this. I just found it to be an interesting snapshot of a particularly chaotic point in the company's history.
> Does anyone ever go back and track the people who were rejected what happened to them?
Almost never, in my experience. Several times I came second in the running and never heard from them again. Contacting them myself didn't work either. Typical shop would rather look thru another hundred randos than revisit one decent candidate.
Not to hijack this sub-thread, but just a general question: how does one deal with unethical 'homework projects'? I've had a case during an application where I was given a 'homework project' that seemed to me very much not like a theoretical scenario, but more like the company wanting me to do free work for an actual real-world issue they had. Have people had situations where you suspected the same? How do you deal with it?
My employer deals with this by compensating candidates for their time. Y'know... with money.
We also don't give them our actual problems as homework tasks. We'll occasionally talk about our real problems with candidates in interviews, but we're very clear about it when we do.
It's not a perfect system. Some candidates will choose to spend longer on the task so that in the follow-up interview they'll have the opportunity to talk about stuff that really shows off their strengths, so their effective hourly compensation for doing it would be quite low. The task is explicitly flexible like this, and we've also hired people who spent _half_ the par time on it (e.g. life circumstances making spare hours hard to come by) and didn't implement much at all, but then were able to confidently answer our questions about the bits they didn't actually implement.
Even if we were to only hire 1 out of every 15 people who get far enough through the pipeline to do the homework task (I don't recall the actual numbers) it costs us _nothing_ to compensate people for their time compared to, e.g., what it would cost us to make a bad hire. So it seems like an obvious thing to do even if only to stop candidates from having to wonder "am I getting screwed here?"
Nearly every worker in the US is covered by an employment agreement that says they have to get pre-approval for moonlighting. It may or may not be legal in their jurisdiction but by paying people to do your interview process you potentially open them up to bad liabilities.
That said it is completely unethical to make people do real work as part of an interview process. We collectively should name and shame any firm that does that.
"Nearly every worker in the US is covered by an employment agreement that says they have to get pre-approval for moonlighting."
This is not true.
I've can think of one employer my entire career that had conditions regarding work outside of my normal work hours.
Granted, if you worked for your competetors and they found out about it you could be fired due to trade secrets or conflicts of interest. I only remember one employer I had where I signed something with binding agreements related to other work (and, as it happens, that was perhaps the worst employer I ever had).
I know some employers in some regions do this, but it is a far cry from affecting "nearly every worker in the US". In fact, I am not aware of any of my friends in tech positions currently being under such an agreement (that is, an agreement requiring pre-approval for outside work). In fact I believe it is illegal under many circumstances in some states for an employer to require it (but don't rely on me for that, conditions/laws change, and I have had no reason to look into it recently). I do remember discussions about it in the not too distant past, however.
I don't know about the US but anecdotally it's been definitely the case for most of my work contracts in Brazil and Sweden for the past 17 years, I believe that only 2-3 startups I worked for didn't have a clause about moonlighting requiring pre-approval from management.
I did do analytical "sample" work for a company that took me two weeks. They were thankful for the work but did not hire me. I published it on my website and they were not very amused.
Agilent is not great either. They interviewed me for 6 month and implemented some of my ideas. (In fact, I got the job, out of 500 applicants, but then they axed the job and hired nobody)
One downside is that to do this properly, you probably should be sending out 1099s for all of those candidates. Depending on the number of open reqs out there, your payroll department may be less than thrilled with this idea. I’ve seen some teams that have tried to sidestep this by paying in gift cards, but that sure does seem to be playing with fire.
You are not required to file a 1099 if the payment is less than $600. Even then, I can't imagine it's that hard to send out an additional 1099 if needed.
Doesn't this discriminates against people who are on visas who can't accept remuneration from anyone other than their current employers..? They then either have to do the same project for free or decline to continue the interview process.
Not only visa, but also while employed one often can't simply be paid by a "competitor" for similar work, that would break the loyality to the current employer which is mandated by contract law and general low.
This is always, always the answer. It's not only the decent thing to do, it ensures you'll get an accurate representation of the quality of the prospective employee's work.
It's pretty simple: the interviewing process goes both ways. The company interviews the candidate, the candidate interviews the company.
If you feel that a request is unethical and it's not what you want in a company, well... you - as a candidate - can terminate the interviewing process.
Simple - you give candidates assignments you've designed solely for interviewing. Ones that you clearly state up front are not real, not based around current project deliverables and commit that the code will ever be used in production.
Then it's no different to the rest of the candidate's time spent interviewing - it's cost of doing business that no one expects to be remunerated for.
The other benefit is that you then keep a benchmark of deliverables by having each candidate complete the same task (you have to change tasks over time as the details of the project get leaked/discussed).
When I give homework projects, they are extracted from a previous part of the interview: What is a personal project that you’ve wanted to do but haven’t had time to do? Then I just ask them to do that. Win / win.
We give new hires a homework project that is directly work related. We don't consider it unethical. It's really a fantastic way to sort out who can deliver. We used to have it be part of the onsite interview - they could sit for a few hours and do it.
I like these because they are not games. If you want something done, you'd ask an employee to do it. So just ask someone to do something you need done.
In our case at least by the time it was something that was part of an interview, it had already been implemented on the business side. Our projects were usually 2 hours tops?
The idea that this is unethical is wild.
We also do paid internships and have folks actually work on stuff that way -> do a good job, pretty good line up for a full time position.
You are misunderstanding. They are saying it is unethical to ask an interviewee to code for free and use that in production. In your case you are not doing that, although if I interviewed with you and saw the same feature added after I submitted the code I would definitely assume you did something unethical.
But is that what is happening? They said it was a work related project. That's exactly what we do.
Why would a place like google even use an interviewees code without careful copyright assignment and work for hire protections (ie, you need to pay someone in USA generally to own their code).
I've found some potential hires are randomly paranoid - and if they start giving you lots of hypothetical disaster / ripoff scenarios early, not worth the hire?
Sometimes companies do illegal things that save them money. I would assume Google wouldn't do that. Not every company is Google.
A 6-person company in New York did that to me. I reviewed the interview project I did for them and their latest product update. I had to ask them a week after the interview if they used my code in production and if so - this is how many hours I worked on it and a fair rate. The CEO emailed me threatening legal action and then called me 10 minutes later apologizing and venmo'd me the amount I stated it was worth.
If your interview process works for you that is great.
But I could totally see why potential hires are paranoid if you are giving them a situation to be paranoid about. I hope you are being super clear with your process and giving assurance you aren't using their code. If it was me, I would show them the code our team wrote at the same time they submit their project and use that as part of the follow up interview.
This happened to me in am interview. It was my first software engineering job, so I was a junior. Wound up solving a problem one of the senior engineers had been struggling with for days. It’s probably what got me the job. Assuming the company is actually hiring for the role I think it’s perfectly fine to do this. It identifies people who can fill in gaps.
We don't use homework projects at my current job (I don't care for them personally and thankfully have a bit of influence in the hiring process), but I've worked places that did. We always took real-world problems we had but had already solved. This has a couple benefits: a) we have full context around edge cases, bugs, historical stuff, etc. that the candidate doesn't; b) we can see pretty quickly when a candidate does something innovative we didn't think of and it's always lead to extremely good discussions in the interview; c) it's very similar to the work they'd be doing day-to-day, so if they have a lackluster resume and/or don't interview well, but shine on this, their odds of an offer skyrocket; d) the company gains nothing from the work so that settles most (but not all) of the ethical objections to it.
I just don't do them and move on to the next company. I've never had an issue landing my next gig. I'll do all day multi round interviews and white board questions. I'm not doing your work or taking a college exam again. Good luck and godspeed is all I can tell them.
Ubiq (YC company) interviewed me with an 8 hour assignment that had ridiculously challenging requirements. It was very fun because there was no way any normal human could do it in 8 hours... but how far could I get? And they paid me for 8 hours working on it.
I felt this was good. I got paid and it was a fun challenge. It also obviously wasn't "actual work," but even if it was, I was getting paid, so why not?
I don't think it would have been out of the question to say I was too busy to dedicate 8 hours either, but in my case I had the time.
I had to solve a real world problem for my current employer in my assignment. However, when they usually say it takes 20% to make a nice demo and 80% to make a the demo production-ready, I'd say the distribution for my assignment was even more uneven. It's just a PoC, a toy, nothing that can really be used. I did not know the tools used, the coding style or anything when I coded the assignment.
I see nothing unethical in it.
3 years later the feature has still not made it to production. I hate that every time it would have made my working day easier :)
I will only ever commit to a homework assignment if the employers' commitment level is on par.
For example, if they want me to do a homework assignment before meeting the hiring manager, no way. In other words, the assignment CANNOT be the first gate otherwise companies will just spray and pray then find the most desperate candidates.
Overall I don't think the total amount was excessive, or maybe just slightly. My experience has always been been technical phone screen or homework problem before onsite, but not both.
The real issue was how chaotic the process was. I had started out talking to an engineer that had picked up my application, but was handed off to recruiter only after the homework portion was done (said recruiter even mentioned that they had just been hired). The two technical phone screens also were two discrete steps (i.e. the second was only scheduled after the success of the first).
That was really the issue: neither I nor them knew how many more steps there were. It wasn't helped by having long periods of radio silence between each step either. But it was painfully obvious that they were scrambling to figure out how to scale their processes, so much so that it still stands out to me 6+ years later.
As a sysadmin, you probably have a number of areas of responsibility. Imagine a time in your career when you had the least amount of sleep, and still tried your hardest to keep everything together. Then you meet someone who gives you the feedback that they enjoyed talking to you, but they noticed that one of those areas is not quite as buttoned up as it should be. Presumably you take that feedback to heart and fix it as quickly as the circumstances allow. 6+ years later, that person goes on HN and takes you through the dirt. Would that seem reasonable to you?
Thanks for calling this out: you're right. I don't think I highlighted properly what I actually wanted to highlight.
I more wanted to emphasize my view as a bystander at a very hectic point of their growth, not actually complain about the experience. It was an interesting experience to me, not a negative one.
I have no hard feelings about it and I'm sure things got sorted out. I certainly have no negative views of any individual I dealt with.
My experience is that not only did they start out as incompetent, they actively downplayed early security issues and outright lied about certain aspects of their products.
I found major, glaring security/architecture issues with their main product early on, and was told by DO staff to go full-disclosure because it was working as designed. When I did so, DO lied about the impact on their blog.
Your issue was exactly, and I mean exactly, what I had in mind when I said “learning experiences” upthread. I followed it closely. Linode painfully learned the scrubbing lesson as well - and DO was in a worse position to learn it at the time, being forced to throw the scrub write load at SSDs that don’t like that type of load. That’s the real issue, honestly, that they probably started noticing SMART warnings from scrubbing by default on SSDs (particularly for cheap instances that turn over often), and took the venture capital approach of taping over it to make the next round by making it opt in.
It’s a tough problem to deal with and not something you’re likely to think of designing such a product from first principles. That’s not an indictment, it’s just fundamental to experience gained doing this stuff (and it’s perilous to get wrong). Filtering unsolicited ARP to prevent domUs from hijacking default gateways was another lesson in blood, and one of the first things we tried within five minutes when we did competitive on DO around launch time (they hadn’t thought of it; it worked).
The whole scrub thing was a red herring: you don't need to scrub anything to not leak data, you just thin provision. It's (mostly) fine if customer data stays on your disks after they delete. It's not fine to give it to the next customer because you don't know how disk abstraction works.
My issue is mainly how they coped with it, which has nothing to do with their (at the time) technical incompetence: they simply lied about the effects.
Competent or not, lying on your corporate blog isn't a good choice. It doesn't take any special skills or training to be honest on your journey from incompetence to competence, all it takes is integrity.
Yeah. Agreed. I unfairly walked back solve to “deal with” on you simply from knowing the considerations that go into bursts of write load every time a customer clicks “delete”.
At the time Linode had a host-side scrub daemon that simply ate LVs as customers deleted. It’s technically simple, but drive longevity and capacity on the host (the user is probably recreating their VM, for example) is where it gets tricky. External considerations. In the end scrubbing is basically killing a few dozen inodes of user data, but the architecture in VPS usually requires you to nuke the whole image, including what is ultimately the host’s 59th copy of Ubuntu. Just managing the iops without annoying neighboring customers is a challenge.
To my cloning point I made at the top of the thread, Linode was certainly no stranger to the reality distortion field when it suited, and I think that’s yet another thing DO copied. That vertical is closer to B2C since you’re usually dealing with individuals, not sales teams, and glossing over stuff is a bit easier than when your counterparty has a better engineering team than you.
Use an abstraction layer between the virtualized block device and the underlying storage system, and only write used blocks to disk, and return zeroes for all unused blocks.
You don't need to wipe, zero, or TRIM anything to not leak customer A data to customer B.
I second that. Over last 6 years I happened to login and see other people dashboard a few times. First time it happened hair raised up on my arms thinking someone else can see my stuff. I simply logout and logged back in. I never manage to delete or do any reckless stuff with other people servers. But I feel what the OP was going thru.
As somebody who followed SliceHost (Jason Seats and Matt Tanase) the predecessor to DigitalOcean, I 100% knew that DigitalOcean was on a trajectory to exit/IPO and be huge early on. Bravo, and congratulations to the entire team.
I was also building a startup piggybacking on hosting providers around 2012 and 2013 and saw the explosive growth. DigitalOcean should be the go-to case study on how to build a developer oriented company. Just like SliceHost and PickledOnion before; the technical guides on setting up LAMP stacks, Wordpress, NGINX, Node.js become resources just as important as Stackoverflow and Serverfault. DO showed up first in Google search results for technical questions.
I actually prefer that companies stay private so they can have more control over their long term outlook rather than slowly moving to quarterly profits and quarterly thinking mindset that haunts most of the industry tech and otherwise. It was great while it lasted, and I guess those early guys are going to make a killing on the options. Congrats to them. Nothing lasts forever.
The best year of my career was at DO in 2014/15. We were growing like a weed. It was hard to appreciate the magnitude of the growth until I compared it to everywhere I've worked since.
It was a heck of a lot of fun. Thanks for bringing me on board.
I love DigitalOcean. The simplicity of their UI, friendly CLI tooling and (as a hosted k8s customer) the frequency of updates to their offerings.
BUT:
1. Their web UI stability is terrible (constantly hangs and errors out)
2. Their hosted DB IOPS aren't competitive with other PaaS providers. There's also no way to scale DB size dynamically without upgrading to a new price tier which is overkill.
3. Their Kubernetes Control Plane scaling is obfuscated. I had to contact support to realize that the control plane nodes are somehow tied to the size of your node pool when you create the cluster? That's not documented anywhere and was the cause of many control plane timeouts.
4. No ALIAS dns record, no way to serve Spaces from an apex domain. (I know this is a vendor specific implementation but it's table stakes imo)
Again, I'm a (mostly) happy DO customer and wish them the best but if they really want to be seen on the same level as AWS or GCP in terms of some parity, they need to include some of these QoL features.
Yep great SEO Strategy and raising Brand awarness. Win-win for people searching. Find out about how to install nginx on Ubuntu and then memorize that next time you need a VM you go that blue website with great tutorials.
As I started learning how to spin up and manage Linux servers and then deploy my own (unmanaged) k8s on them, half of my bookmarks are DO documentation.
I agree, I had limited knowledge on a few things and was able to get it all sorted with their guides. I think they offered an exchange of credits for tutorials or something. My favorite part of their guides, the drop downs to pick your version. Google can give me results to a simple ubuntu question that is ten years old, they had the option to pick which version you were using to get you the correct guides. Brilliant
I hold up this part of DO as one of the single best uses of the amazing tech writers our industry has, under appreciated and under paid, to punch way above their weight in a crowded market.
I'm sure their own documentation is good, but certainly early on the tutorials always seemed to be a mess. Lots of similar articles written by different people with different structures and advice, without referencing other related tutorials. There seemed to be very little in the way of editorial control, but maybe that's better now. It was a good, cheap content strategy for SEO, but not a genuinely good resource.
Thats because a lot of it was lifted directly from the web. They paid people a stupid price per article to either write articles or copy paste documentation online while barely vetting it, only taking it down when people took the time to send in reports to staff. I'm sure things have improved since then.
Having 5 people install slightly different varations of a stack provided so much value it was and still is the best resource when installing a new stack.
I actually think their UI is too form over function. One particular annoyance I had recently was with the DNS zone editor.
It truncates both the record contents and the domain name, which means that if you've got a bunch of long domains with TXT records, telling them apart means copying the contents into a text editor to see the truncated text.
I agree. I think their UI is actually a little too creative at times. It has lots of whitespace and a single color and that means people will say "clean", "minimal", "modern", "artsy fartsy".
But I agree, that with something that coins itself "the developer cloud" it should provide critical information first. It is ok to kill some whitespace or put some monospaced fonts in there. You aren't trying to sell me an iPhone, I am trying to manage a server.
Route53 on AWS is what I have been using for domain management and I like it because it doesn't mess around with cute UIs. It is strictly function. Instead of form fields for each TTL line and host line like other sites do, you just get a big multi-line text box and you use spaces and line breaks (gasp) to create your dns listings.
Yes, you read that right, they assign a /124 if you enable IPv6. There is no provision for getting anything larger. All configuration is totally static as well. It's really incredible how botched the setup is, and it has been like this for years.
If you're wondering if you're sharing the same /64 with an entire datacenter worth of droplets, the answer is of course.
This -- this has me a bit frustrated and I'll eventually email them about it.
I haven't contacted them about it but that makes sense sadly. So they're getting a /64 from their upstream and just sharing that out to everyone. This isn't IPv4, they need to make this right.
Vultr, OTOH, gives each HOST a /64. I've been doing most of my work on Vultr right now so I can use more than 16 IPs on a single host. I was/am hoping that providers would standardize somewhere in the middle: Every customer account in each datacenter gets a /64. I wouldn't mind all of my hosts having their own /64 to subdivide.
The smallest allocation a customer should ever get with IPv6 is a /64. You break a lot of stuff when you go smaller. DigitalOcean's approach is kind of like sticking an entire datacenter behind a single IPv4 address.
It's not like it is hard to get IPv6 address space either. All they have to do is ask. A VPS provider like DO can get a /32 easy peasy.
I don't know if I'm doing something wrong but am I right that a droplet firewall doesn't have any ability to create source exceptions to lock an open port down to a singular IP. Seems like a basic enough feature but I can't seem to find it.
Not being on the same level of feature parity with AWS/GCP can be considered a feature. Not everyone wants to drown in service orchestration and configs
100% - it's clear how everything works because they've pared down to the essentials. For their target market of small dev teams who just need to ship code, it's a godsend.
Maybe we're talking about different things, docs here[0]
> You cannot assign a Floating IP to a Droplet created from a custom image.
Edit: I wanted to double-check because this would make my life so much easier, but no, I still can't use Floating IP with Custom Images. I imported a FlatcarLinux image from a URL and created a droplet; once the droplet launched, I still see this:
> Floating IP: Disabled
> Public IPv6 Address: Not available with custom images.
Yeah, Snapshots work quite well, you can even transfer them between accounts and copy them between regions in the same account, but it's a "transfer" meant to be used as a migration mechanism, not distribution.
It seems like Amazon et al have full-fledged SDNs while DigitalOcean is a little more primitive. Even customer-level VLANs are new. Prior to that, "private networking" was just a way to save on transit quota; you shared a LAN with everyone else in the datacenter.
Speculating and recalling old memories here, but: floating IP is implemented as an additional IP on an existing interface. DHCP can't assign multiple IPs for the same MAC. So there's got to be some DigitalOcean magic in the base images to figure out that a floating IP is assigned and configure the interface for it. You might be able to replicate this yourself, they just don't want to be responsible for support.
Perhaps over time they will grow a more complex network fabric that will enable floating IPs to be their own DHCP-enabled interfaces straight on VMs, instead of the "dumb" forwarding rules they appear to be today. But also perhaps the lack of such things is why DigitalOcean is cheaper than Amazon.
Agree on all points (and definitely, a mostly happy customer too).
(3) was the cause of numerous production incidents for us. We had to contact support to have it scaled up, and sometimes they'd take up to 3 working days to get back to us. Happily paid more for AWS to get better support, and stability.
Nothing is perfect. We use DigitalOcean Droplets because they're more bang-for-buck than AWS EC2 instances, especially if you're doing a lot of disk IO. However, even though it's more expensive we use AWS S3 instead of DigitalOcean Spaces because it's faster, more reliable, and replicated automatically. I wrote about these decisions recently here:
Their support is also very lacking, multi thousand a month spend with them buy tickets still take hours. But support seems the hardest thing to scale especially with the number of tickets they get about things not their fault.
I'm _really_ disappointed with the managed DB service. The "75 simultaneous connections per gigabyte of usable memory" limitation means cost wise it's cheaper just to spin up a 'real' database droplet.
FWIW, if you’re using 75+ connections, it’s probably a good idea to add some kind of Postgres proxy in the middle like pgbouncer. Postgres doesn’t handle lots of connections well because each connection forks off its own full OS process, and the performance can degrade noticeably when you have too many connections to Postgres as a result.
Really like the direction DO is going. Which is something in between millions of small VPS firm and giant like AWS. It also forced Linode to React. And proved the space in Cloud Hosting exist, which company like Vultr and UpCloud are now joining. And I think their features are finally close to hitting the perfect spot. With all the essential like DBaS, Object Storage and K8s.
They have also ( finally ) introduced AMD EPYC CPU, ( no more explicit mention of Intel CPU on their homepage ). Which is good for their margin since price / core count are cheaper.
So DO today is something I would like as an AWS competitor in 2016. And they are finally here. I am not sure if there are any other low hanging fruit at all. Or what features I would want for a company competing at this space. It will be things that people dont see, like CPU performance, SSD Performance, Services Reliability, Interconnect, Better Network, More Locations, security improvement....
There are only two suggestions I could think on top of my head right now.
First is the pricing structure and list. It is getting very messy and AWS like. SSD Storage variant should be a simple calculable option, not an extra line on pricing table that tries to bombard me with extra information. And I know Basic Droplet are burst-able CPU resources, but not showing they are vCPU ( a Single Thread ) and calling them CPU doesn't really rhymes with me.
Second is Full BareMetal Monthly option. Something like vultr [1] is offering. I am not sure if below $100 is feasible. Basically it should be something that offer much better price / performance at a monthly payment that pushes people with low spending to metal for baseline load with headroom as insurance. Pushing up Average User Spending. It should also be attractive to small business. ( Although arguably Linode and Vultr has yet to expand or launch Metal seems to suggest otherwise )
I'm a fan of DO too. In terms of "what's missing?", how about the equivalent of AWS Lambda@Edge or Cloudfront EdgeWorkers? ie, a NodeJS runtime as close to the end user as the nearest CDN edge.
The problem with Runtime at Edge is that they are extremely engineering intensive, given it isn't just hosting but the whole software stack involved. ( See CloudFlare Workers for Example ) Which is sort of outside the sweet spot that DO is currently enjoying or operating in. I am not sure if it is a sound strategic decision to persuade. Although I would certainly love to see competition in that area.
Well yeah, if you care enough about performance that it matters whether the "server" runtime is at the data center or the edge, it's kind of a given it won't all be completely automagic. But it is, thankfully, becoming an increasingly straightforward path to tread.
We were going to call it that, bought the domain names and everything, but a competitor registered a trademark on the name the week before we launched. Oops.
We are an “emerging growth company” as defined in the Jumpstart Our Business Startups Act of 2012, or the JOBS Act. We may take advantage of certain exemptions from various public company reporting requirements, including not being required to have our internal controls over financial reporting audited by our independent registered public accounting firm under Section 404 of the Sarbanes-Oxley Act of 2002, or the Sarbanes-Oxley Act, reduced disclosure obligations regarding executive compensation in our periodic reports and proxy statements and exemptions from the requirements of holding a non-binding advisory vote on executive compensation and any golden parachute payments. We may take advantage of these exemptions for up to five years or until we are no longer an emerging growth company, whichever is earlier. In addition, the JOBS Act provides that an “emerging growth company” can delay adopting new or revised accounting standards until those standards apply to private companies. We have elected to take advantage of certain of the reduced disclosure obligations in the registration statement of which this prospectus is a part and may elect to take advantage of other reduced reporting requirements in future filings. As a result, the information that we provide to our stockholders may be different than you might receive from other public reporting companies in which you hold equity interests.
Wow, I didn't know about this. Shouldn't financial reporting always be transparent?
The short answer is theoretically, yes, but in practice, it's not always practical to have transparent financial reporting.
For context, financial reporting is a tradeoff between cost and effectiveness. Whenever you're reading audited financial statements, you're reading an accounting professional's opinion which would be reasonable given a certain level of constraints. In theory, auditors could audit every facet of an organization and obtain 99.99% assurance, but the financial cost of doing so typically doesn't make sense for the company nor shareholders.
Of the reduced disclosures, the most significant is not having their internal controls audited. For a big company, this is a red flag because the financial accounts are only reasonable if you also have reasonable assurance that there are controls in place to prevent fraud and that they're working effectively.
But for smaller companies where most of the ownership is usually owned by founder-workers, employees, or early investors who are monitoring it on the ground level, there aren't many benefits from increased reporting over internal controls because if they are committing fraud, they'd mostly be defrauding themselves! That, combined with the fact that most early stage companies are already resource-constrained, makes regulators a bit more lenient because they assume investors/employees know what they're getting themselves into.
Now, when a company decides to go public, they need some time to adopt best practices and comply with broader regulations. That takes time, so regulators give them a few years to get the personnel and processes in place without penalizing them. But to cover their bases, they're required to make disclosures like above, so that early investors buying into the IPO know that they won't have similar levels of assurance about the financials for a few years.
> A company qualifies as an emerging growth company if it has total annual gross revenues of less than $1.07 billion during its most recently completed fiscal year and, as of December 8, 2011, had not sold common equity securities under a registration statement. A company continues to be an emerging growth company for the first five fiscal years after it completes an IPO, unless one of the following occurs:
> - its total annual gross revenues are $1.07 billion or more
> - it has issued more than $1 billion in non-convertible debt in the past three years or
> - it becomes a “large accelerated filer,” as defined in Exchange Act Rule 12b-2
2020's gross revenue was 318m growing at 50-60m yoy from prior years. So, unless that growth is somehow compounding, the 5 years post-IPO is the most likely outcome.
The biggest implications are relaxed requirements around explaining executive compensation, and that financial control auditing (SOX-compliance) is not required.
It's not necessarily a bad thing for investors, but a trade-off. It means the company can focus more on growth and less elsewhere.
They've accrued up losses of $43M, $40M and $35M in the 3 preceding years. Funny how that's considered financially healthy in the world we're living right now.
I think it's good to point this out and think a little about what "health" means here. Looking at their consolidated financial data on page 11, you can see they spend about $1 providing a service someone pays $2 for. Then all the rest of it gets eaten up by salary, r&d and marketing.
It feels like a very different situation than, say, Uber - where there's a question about if Uber can balance driver payouts and customer fees in a way where they make money. DO, instead, is already turning $1 in goods into $2 in revenue and they're just looking to grow enough so that $1 in potential profit covers everything.
25% revenue growth past 2 years and losses have remained constant.
If this wasn't hard tech, I would be worried, but this is hard capital intensive stuff and looks to me like they are doing fine.
If in 10 years there ends up being 1 independent cloud company, that would be a pretty amazing business. DO has a good a shot as any (only thing I can think off is AWS spins out). DO is so good that there I can't think of a worst case scenario being anything other than one of the big 3 buys them. Best case is the moon.
So long as the investment into growth (sales and marketing) is paid back through the cashflow of customers during their lifetime, the losses are tolerable.
A provider like DO expects to have customers for many years. At 50% gross margins, $100 spent to acquire a $50/yr customer will be returned in four years. That may be an acceptable trade off.
Depends, what are they spending the money on? Are they wasting it on user acquisition or are they using it to build more datacenters? Are their losses projected to grow or are they staying the same while increasing revenue? Are customers actually using the service or are they just being lured in through loss leaders?
One should probably create a personal checklist to see whether a company is doing well or not and then just make decisions based on the checklist.
The business is making money, they are just spending it growing the business. Most tech companies that are still in the growth stage will lose money because they have to spend a lot on R&D and keeping employees at the company with compensation. The tech industry is highly competitive and you have to spend a lot to stay competitive.
I ran into issues several years ago where the IP of my DigitalOcean droplet always seemed to be on block lists or banned for abuse from various services. As a result it made it impossible to use as a VPN server or mail server. The early ability to spin up and charge hourly rates made them good targets for spammers compared to companies like Linode, who at the time didn't have hourly rates.
I have to imagine things have gotten better, but it did taint my view of DO as more of a dev playing field than a reliable hosting provider.
It hasn't gotten better. I had to plead with a postmaster at a large ISP to unblock my droplet because they banned all DO's IPs after they were refusing to manage the way spammers were abusing the service. Not a "we're sending to spam" ban but a hard 550 connection refused. The postmaster suggested I do some Googling on the topic and I was shocked at what I found, so I can see why they had to throw their hands up and block the lot.
I appreciate it's not easy problem but it's clear the community felt they were being ignored and DigitalOcean did nothing to try and limit the spread. For example, moving to a model like other providers have where you need to request permission to send mail from support.
EDIT: Just to clarify, AWS has a lot of IPs, over 100 million [1]. Let’s just speculate that 1/50th might be in an overall pool for customer allocation. Do you really think the majority of 2M IPs would come back blacklisted in some form? I haven’t dug in to this to actually see if this is the case, but it would seem laughable to even begin to think that this might be the case..
I think the wording should be changed. I would agree that all their EC2's and lightsail instances would be in the RBL/RSL's. AWS have an email service specifically for this use case.
They come and go get listed and delisted so you may find your own nodes to be hit and miss. That is for the well behaved RBL's and RSL's. Spamcop, Spamhaus, Uceprotect will expire nodes and networks after a period of no reported spam. There are a few others that will leave things listed for a long time. I am only basing my experience on those three. The less costly a VPS provider, the more often they get listed. Ramnode for example is almost always blocked. If you have long running nodes, it is unlikely you will find them blocked outside of Uceprotect level-2.
That's pretty much me too. I briefly thought they had a major Kubernetes fail recently, but they immediately diagnosed the issue and got back to us...We'd shot ourselves in the foot. I love DigitalOcean.
I might have mentioned this before, but DigitalOcean's support is (or was, before I stopped being their customer) absolutely abysmal.
I had a default card set up for auto billing. That card expired. I went to remove the default card -- you cannot remove a default card. Okay. I added a new card and tried to set it as default. It didn't take. (Refreshing the browser reverted the change.)
So I tried another browser, another computer, etc. The change never took. Finally, I contacted their support. It took a ton of back and forth with people who had dubious language comprehension before I finally convinced them anything other than user error was going on... but then, instead of fixing it, they demanded I take a video and upload it for them to prove it was real!
I did, and I paid the invoice manually. The bug was eventually fixed after a month or two, but it was a pain until that happened.
I found another provider who is cheaper for the resources I need and has better customer service by miles. If I can't get you to even take my money in an easy and reliable way, I can't trust you to operate anything I rely on.
My support issue was that I had a dead box and needed it to be available again. Their system thought it was up, but I could not reach it. There was no live support, just submit a ticket and pray. They need real-time support. I’m fine paying for it. Hell, bill me per minute, but I needed eyes in the issue. Instead of waiting for some unknown time for a support response, I deleted my stuff and rebuilt everything. If this were backing a commercial offering of mine, this would be unacceptable.
This may be a dumb question but why would you put anything on a server at DO that you need 5 nines out of? DO is all about the price and mediocre tools and machine that you get for it. I use it for personal/fun stuff but I'd never put my business on it.
Sure, you probably don't want to run your business on it; we agree. I think it is a missing opportunity for DO to allow growing smaller projects to continue to stay on their platform.
I moved to BuyVM, https://buyvm.net/. I had already been using their services for a few years, I just expanded on what I already had.
They are not as large an enterprise as DigitalOcean, by a long shot, but the upside of that is you can speak with the owner and his small staff personally and get personal service. They made me a very good deal.
I also know from their Discord that they have some conservative political leanings, so if that's a thing that matters to you either way, it's there. I don't find it to be represented in their behavior towards customers or who they choose to host, so it was acceptable to me even though I feel differently.
They are really the perfect size for me: big enough for multiple locations, R&D, and support coverage; but small enough so that you personally interact with the team, and you have trust that a random DMCA isn't going to end up with your site nuked.
The privacy-first stance (including allowing Tor exit nodes! sorely needed on independent infra!), and general willingness to help you ad-hoc (e.g. you wanna announce your own IP range? they'll do it, charge you a fair price for the work).
On the other hand, I woke up one day with my DigitalOcean droplet suspended and account banned, and support absolutely refuses to tell me why. My only suspicion is perhaps because I used the GitHub Student Pack even after I was no longer a student.
It blows my mind that people will hang out in a Discord for a VPS provider and express their political opinions. Communities form around anything nowadays.
I'm not for or against BuyVM. I'm just surprised that nowadays people will hang out in a VPS provider's Discord. Like if Wawa had a Discord and people decided to hang out there. I just don't get it.
I think people would totally do that for Wawa. I know some people who are crazy into the whole Wawa vs Sheetz debate.
There's a whole community around "low-end" VPS and server hosting, and BuyVM's staff have historically been active figures in that community. The Discord is an offshoot of that. It's not all that unusual; a lot of companies do community outreach by running communities for people who do things related to their services. Digital Ocean themselves do this in another form with their documentation libraries.
BuyVM's brand relies on the perception that they are laid-back, pro-free-speech, friendly, and helpful. The Discord is a way of broadcasting that. Now, personally - I think that attracts some people I wouldn't prefer to attract. But I know they've drawn lines in the past (banning overtly anti-Semitic people, for instance), so, fair enough. Their pro-speech attitude helped me when a site I ran was subjected to bogus legal threats from an unhinged former user.
BuyVM's wiki has things in it like https://wiki.buyvm.net/doku.php/kvm#pagefileswap_disclaimer, which, without any numbers behind it would make me extremely uneasy to use the service. Other cheap hosts just put hard IO limits at the VM host level.
BuyVM is a team run by humans who do care about every customer. As long as you're doing something malicious-y (like crypto mining, exploit testing random websites, distributing malware or a phishing site, etc), they will work with you.
> As long as you're doing something malicious-y (like crypto mining, exploit testing random websites, distributing malware or a phishing site, etc), they will work with you.
Is this an extension of the Russian Business Network, or did you miss a negation in there? :-)
Yeah. There are definitely people on their Discord I'd call extreme, but none of those people work for or speak for the company. They aren't people I would choose to associate with, but it's a free country.
I thought about it quite a bit, actually. But there is a line there and they haven't crossed it.
I use Vultr FWIW, I got $50 in credit way back when, and they're stable, and their support was responsive, if a bit unhelpful, regarding a firewall issue.
A small personal project of mine was flagged as compromised or something and taken down. Support couldn’t tell me why, I suspect it had something to do with updating regularly through git? To this day I’ll never know though, moved it to an aws ec2 instance.
Maybe they have support tiers? Every time I created a support ticket I got a response within an hour from a technical person. Their support responds much faster than AWS in my experiences.
The biggest threat to DigitalOcean is that AWS, Google, and Microsoft are all going vertical by designing and manufacturing their own chips. Over times they'll be able to get hardware for cheaper, and also more specialized hardware that uses less power.
DigitalOcean will have to buy CPUs from Intel or Nvidia/ARM at a higher cost, and eventually maybe even from AWS or Microsoft, essentially giving money to their competition.
If DigitalOcean doesn't get into semiconductors quickly, the only two logical outcomes are to either go bust, or to be acquired by a major cloud provider as a low-cost branch, like airlines do.
I think there's a lot of doubt whether, in the long run, doing your own hardware makes sense for a cloud provider.
If you work yourself into a performance dead-end, you can spend billions and end up with a chip that isn't particularly competitive. Whereas if your supplier does that, you can switch suppliers.
You could spend $X billion, end up with a chip that isn't particularly competitive and just have to eat those billions.
You bring up airlines... airlines don't build their own airplanes.
Copying a reply [1] I did on HN some time ago. Which should add some perspective on the point being discussed.
AWS are estimated to be ~50% of HyperScalers.
HyperScalers are estimated to be 50% of Server and Cloud Business.
HyperScalers are expanding at a rate faster than other market.
HyperScaler expanding trend are not projected to be slowing down anytime soon.
AWS intends to have all of their own workload and SaaS product running on Graviton / ARM. ( While still providing x86 services to those who needs it )
Google and Microsoft are already gearing up their own ARM offering. Partly confirmed by Marvell's exit of ARM Server.
>The problem is single core Arm performance outside of Apple chips isn’t there.
Cloud computing charges per vCPU. On all current x86 instances, that is one hyper-thread. On AWS Graviton, vCPU = Actual CPU Core. There are plenty of workloads, and large customers like Twitter and Pinterest has tested and shown AWS Graviton 2 vCPU perform better than x86. All while being 30% cheaper. At the end of the day, it is workload / dollars that matters on Cloud computing. And right now in lots of applications Graviton 2 are winning, and in some cases by large margin.
If AWS sell 50% of their services with ARM in 5 years time, that is 25% of Cloud Business Alone. Since it offer a huge competitive advantage Google and Microsoft has no other choice but to join the race. And then there will be enough of a market force for Qualcomm, or may be Marvell to Fab a commodity ARM Server part for the rest of the market.
Which is why I was extremely worried about Intel. (Half of) The lucrative Server market is basically gone. ( And I haven't factored in AMD yet ) 5 years in Tech hardware is basically 1-2 cycles. And there is nothing on Intel's roadmap that shown they have the chance to compete apart from marketing and sales tactics. Which still goes a long way if I have to be honest, but not sustainable in long term. It is more of a delaying tactics. Along with a CEO that despite trying very hard, had no experience in market and product business. Luckily that is about to change.
Evaluating ARM switch takes time, Software preparation takes time, and more importantly, getting wafer from TSMC takes time as demand from all market are exceeding expectations. But all of them are already in motion, and if these are the kind of response you get from Graviton 2, imagine Graviton 3.
Although the difference with airlines is nobody is pooling the wealth of the world into airlines, investors think tech is going to take over everything. Your cloud providers are being valued like they are going to either run the world at some point or get some compensation for being dismantled by international policy. And when your investors think this way, and you make as much money as these companies do, you might as well make your own chips.
Plus you have to provide most, or maybe all, of the support for software. Not just making compilers and related dev tools, but also building and testing of pre-compiled binaries for popular OSS.
That's fine for a few major pieces of OSS, and for internal, cloud-only software (Redshift, BigQuery), but I don't think that concept scales well, or is particularly quick to adapt to market shifts.
Also, to pile on, for any company that runs a real sales process, Redshift and BigQuery are losing to Snowflake. So the argument for custom chips doesn't seem that compelling.
It might be the megacorporate equivalent of "strategic assets" for nation-states. Keep a chip division going and spun up, if only to have a core competency as a backstop.
> the only two logical outcomes are to either go bust, or to be acquired by a major cloud provider
Or just continue to build a profitable, cash flow positive business that continues to grow for years and years to come. Business outcomes are often not binary.
For GCP, it depends. GCP doesn't have a Nitro competitor, but they offer Tensor Processing Units (TPUs). TPUs are more of a competitor to nVidia chips, but they are still custom silicon in the sense of the GP comment.
GCP, AWS, and Azure seem to all be diverging in their specialization. GCP is focusing more on AI/ML/Big Data offerings while AWS is more security-aware, and Azure is more hybrid-cloud focused.
GCP feels the most R&Dish, with the K8s and ML innovations. AWS seems to just be pushing on every front very well. Azure feels as though it's catching up, but a bit more sales led/engineering later than the others.
“Because silicon is a foundational building block for technology, we’re continuing to invest in our own capabilities in areas like design, manufacturing and tools, while also fostering and strengthening partnerships with a wide range of chip providers,” says Microsoft’s communications chief Frank Shaw.
Thank you DO for offering hosting for us in Africa who need a reliable hosting and fixed rates for the services so that we can plan ahead. Also the tutorials are a must have! Like today I used Django and DO Space and everything was smooth.
I've been using DO for around 6-7 years now. Really happy with the services overall. I hope nothing changes after IPO'ing.
The only complaint I have with them is DO Spaces. This service seems to have issues related to degraded quality pretty often and there's so many horror stories of load times taking hundreds of milliseconds to serve content from their CDN if you Google around for folks using Spaces. I'm looking forward to the day where Spaces is as good as S3 because DO's offering of Spaces includes not only an object store but a free CDN on top. Seems like a good deal for $5 bucks a month if it were dependable.
It's weird because every other service I use by DO has been top notch.
Yep, same experience with Spaces here. That's why we use Spaces for backups only, since it's very affordable and backups don't need millisecond latency.
Excellent growth, fair margins. Probably worth about $3bn. If it IPOs at less than $5bn it's probably worth picking up. Long term digitalocean will struggle to maintain its margins when competing with Azure and AWS on one side, and Cloudflare edge computing on the other side, so I don't think it can command the same kind of premium we've seen from other tech IPOs.
A big red flag is that 570,000 customers bring in only 357m in rev. That's $50 a month for the average customer. That's way too low.
I saw this as an advantage though. They do not have whale customers bringing a significant % of their revenue. I see lower risk here
"Our average revenue per customer (which we refer to as ARPU) has increased significantly, from $35.97 in 2018 to $40.16 in 2019 to $47.78 in 2020. We have no material customer concentration as our top 25 customers made up 11%, 10% and 9% of our revenue in 2018, 2019 and 2020, respectively."
> That's $50 a month for the average customer. That's way too low.
I vaguely remember a talk I had with an employee there was a time when I think the average user was paying only $20/mo, which at the time, I had a $250-500/mo bill depending on load.
This was before they even had a load balancer product (so 2012-2013-ish?), and I was running a fairly unsuccessful managed WordPress offering that had dynamic scaling based on the reported metrics.
The infrastructure was just a big server for MySql, and a tiny server for the orchestrator. Then it would use the DO API to spin up new servers off a snapshot, ssh into them, and run the setup script to point to the right Wordpress instance in the database, then add it to the nginx reverse proxy list for the domain.
The whole thing could scale up or down in 30 seconds. Sure it isn't as fast as AWS or Azure could scale, but we were in control, in code, before containerizing was even really a thing.
I wonder what the distribution of revenue per customer looks like. I bet (hope?) that at the low end there are a lot of consumers who set up a $5 droplet to tinker around for a few months without doing much more, and that the revenue figures are propped up by a small group of bigger customers spending much more.
Believe it or not, $50 is great for a "self service" company like this. No enterprise sales teams pushing complicated billing or contracts, just customers (mostly devs/startups) choosing what they need without endless upsells. Looking at churn and seeing that per-customer number go up over time is the real tell here.
A newsletter about SAAS called CloudedJudgement sent this in January:
> SaaS businesses are valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue. In the buckets below I consider high growth >30% projected NTM growth, mid growth 15%-30% and low growth <15%
So if DO is mid growth, you could use 17x as the EV / NTM multiple. So if 2021 revenue will be 320mm + 25% = 397mm NTM * 17x = 6749mm EV.
Total debt is 263mm so best (rough) guess at equity market cap is 6749mm - 263mm debt + 100mm cash
In today's frothy market, do you think that they'd only be at $3B? Not baiting at all, I'm curious on how you'd break this down as it's always an arcane art.
I mainly wonder as companies like C3.ai are above a $10B market cap on 50% or less of DO's revenue. And C3 (just to extend the example) is carrying lower margin services revenue as well. Growth is somewhat different, granted.
Not an internet business per say, mostly used in the context of processing and storing data + internal apps.
Do is great value, although not the cheapest. They need to double down on the useful items, maybe go a bit offroad in what they offer, lest they might be a in rough spot.
I don't see why you would use DO it's vastly inferior to every cloud provider, the only reason people use DO is for the price, it's cheap compare to AWS, Google or Azure.
That's okay, the comments can prove illuminating. People like the documentation. People like the simplicity. People like the UX. And more. Give the comments a read.
If you value understanding how people make purchasing decisions, this is an opportunity to get some insight because as you can see, there's a lot more to it than what you were able to enumerate.
Seems you might be a software engineer. This should interest you because an engineer that understand the business end of things is immensely more valuable than a heads-down coder in the majority of tech companies.
For some workloads, it's also much faster for similar money.
I looked into migrating my app to Azure, to get similar performance to what I have in DO was over 3x the monthly cost. Worked with Azure support to benchmark and try and come up with a plan to migrate, and the support rep ended by saying essentially "don't think we can get close to your current performance for near that price".
I used to work at DigitalOcean. Can confirm we didn't think a lot about Linode. We saw AWS as the real competitor. If there was someone in the VPS market we respected and tried to emulate it would have been OVH, not Linode.
I agree generally, but Lightsail is just obvious hot garbage. The CPU throttling is crazy aggressive, and you'll see it after 30 minutes of use, for all but the 2 most expensive of the 7 total plans.
I don't see how they could be maintaining a decent customer base with that setup, but I guess marketing is powerful.
I use Lightsail mostly to dodge Route 53 fees, since it comes with DNS management and I have a lot of hosted zones. However I'm also not a business and couldn't care less about the CPU throttling.
I really want to love DO, but I just can't. 2 main pain points I had from working with it for ~3 month:
1. Spaces keys. There is no way to fine tune access rights. Anyone with spaces key can access any space in the organization and read/write to it. I trust people I work with, but there is always room for a mistake, so even a small chance that someone can accidentally nuke our production space makes me nervous.
2. Something from yesterday: we use new DO apps to deploy a static web app. Yesterday I started to get random 404s for some of the assets, so app become unusable. My colleague in Argentina had same issues, but for different assets. We are lucky that it was a staging app, but imagine it was a production app and that would happen over the weekend? How do you even detect that? Run uptime monitor form dozen different locations?
Completely unrelated to DO, but for 2, yes? You should absolutely do something like that if an app needs to be accessible worldwide. There are many reasons an app might become unavailable for different parts of the world, especially if you're reliant on a CDN since POPs can fail or become unavailable in weird and silent ways. Something like Thousandeyes is good for this.
Would be interesting to see App Platform growth numbers. Lot's of competition: Heroku, Netlify, Fly.io, App Engine, Cloud Run, etc. But they may have hit the sweet spot in terms of features and pricing.
Be a giant investor, ie a bank. IPOs are there for banks to be able to buy up shares at a low price, then dump it when retail investors get their turn to run it up before banks sell off.
being an accredited investor doesn’t magically get you IPO shares. they are very scarce and almost completely allocated to high volume institutions (you might get lucky if you have a very large relationship with one such institution but that’s harder and harder these days)
Why isn't it regulated in the US? In India everybody gets a chance to buy in IPOs, it's randomly assigned regardless of how much you applied for. Promoters and institutional investors are even supposed to hold it for 9 months before they can sell it in the market.
all decisions in the US are driven by dollars, and the guidelines are usually made by those that have a lot of dollars. so they create circular situations to benefit themselves.
You can sometimes buy pre-ipo shares on EquityZen or similar platforms if you meet certain standards. It's all very thinly offered though, and sometimes they bundle offers.
I find this really exciting, since DO has one of the only managed PGSQL + K8S combos that seems like it could compete with Google and AWS. Always nice to have more competition in this space. With IPO funding they can probably really compete. They'll probably have their own data centers soon.
been a fan of DigitalOcean for years. Somehow they were able to bring a focus on usability to cloud offerings like no other provider could.
Most recent thing that made my life 100x easier is their Apps platform with direct Github integrations for deployment of react apps removing any need to learn CI/CD stuff which I dont want to waste time on in early alpha startup phase :)
I interviewed at DO a few years ago. I don't want to share the details, but it was the most annoying, unprofessional and disorganized set of interviews that I have ever witnessed in my life (I did ~140-150 interviews in my life for ~25 companies). The best part is that I was introduced to the company by one of their board members.
Since then I kept thinking that I really liked the product, but I would expect them to fail as a company.
Maybe DO's IPO proves that I was wrong, or maybe that they will be economically successful despite what I have witnessed.
edit: if you're going to downvote at least let me know why. I hold HN on high standards.
DO is a fantastic value proposition.
Reminds me when I was gathering cryptocurrency prices from exchanges and needed the lowest delay possible so an algorithm could protect me from pump-and-dumps.
In my testings, a $5/month machine from Digital Ocean was the fastest to fetch prices from Bittrex. I got prices within 0.2 milliseconds (yes, under 1 millisecond). It was unbeliavable.
The algo failed miserably ofc. Mostly because Exchanges price API tend to be unstable. Often giving stale or no data.
I bought my first DigitalOcean VPS for $5/month in high school. The same one is still running today, around 4 years later. One of the best investments I've ever made, allowing me to self-host a ton of my side projects, Discord bots, and anything else I might want.
I've barely used half of my storage and my CPU/RAM is consistently sitting at 30-40% usage. Incredibly affordable for students like me. Plus, their interaction with students is fantastic (they throw free credits at us all the time).
Please add GPUs! Google cloud and AWS's interfaces are hard to use. I recently discovered vast.ai which is great but cheap DO GPUs would be more trusted/professional (even if they only had one option).
I used Nanobox quite heavily before DO bought them.
But was a bit surprised when I saw the below, does this mean that they only payed $3,544 for the company? Or is there more to it than that?
> On April 4, 2019, the Company acquired 100% of the outstanding equity of Nanobox, Inc., a Delaware corporation (“Nanobox”), a deployment and management platform provider for cloud infrastructure. The final purchase price for Nanobox was $3,544 and the acquisition has been accounted for as a business combination.
I love DigitalOcean and would prefer to build my new project on their platform. Unfortunately, the one and only thing that is keeping me from doing so is their support. It's great, don't get me wrong. Whenever I've contacted them I've gotten helpful responses.
However, for this particular project I feel like it is important to be able to get a human on a phone when I need support. If they offered that, I'd move away from AWS in a heartbeat.
but... have you ever gotten anyone useful and able to do anything for you on AWS exorbitantly priced support?
in almost a decade, i havent. actually gotten more done tech related by calling the billing support, then i have through calling paid technical support.
That said all my experiences with DO support were much worse.
Big fan of DO from start. Intutive UI, cheap resources, never had issue with infra. And great knowledge base. But the fact I loved most is a straight billing. There are never surprises at end of month.
Sadly all mistakes done by AWS are picked up by other big vendors like GCP,Azure. Cloud vendors are more focused on adding as many features as possible. Who cares about user experience ? We got covered by certifications.
Anyone else find it interesting that their revenue graph only starts in 2017? At what I’d assume to be a pretty low number given how long they’ve been around and how explosive the VM world has grown? Seems like if they hadn’t taken all that VC capital there’s no way they could have afforded any of their growth.
It worried me that they’re only at a couple hundred MM in a self-proclaimed 116B+ opportunity after 10 years.
It worried me that they’re only at a couple hundred MM in a self-proclaimed 116B+ opportunity after 10 years.
Yes. How do they grow further? DO will be compared to AWS and Azure in the market, and there's a good chance their stock will be punished as a result. Plus, if DO actually makes in-roads in market share, AWS is a beast that, if it wants, can crush them on pricing and features for long enough to kill them.
They have an almost perfect, managed K8S offering.
Major issue that I am facing with them currently is their managed load balancer offering. They have a cap of 40k Maximum Simultaneous Connections / second. I am mostly sure that they are not measuring it correctly.
they have random resource limits which don't make sense.
Also I think AWS lightsail is a competition to DO, but AWS doesn't seem to be promoting it that much.
Thank you Digital Ocean! I've been a customer since the early days and had just a few problems, but they were quick to answer and address. I love the simplicity of their platform and will continue to use their services!
I have a $20pm droplet with ubuntu with auto updates (Including major release) from 2014. So it has done 14->18. DB with web app. I have never had anything more stable since.
Have been a user and fan of DO for years. Their product offering doesn't grow as fast as some cloud providers, but they are still my favorite for just about everything.
Public companies are when private companies have grown to the point that the leaders don't know what to do with it and can't figure out how to compete any longer.
I wonder how all the leftist activist employees that were shown the door feel after leaving six months ago. Bitcoin's meteoric rise, Coinbase market cap at the level that it is, IPO, etc.
Coinbase reduced risk by kicking out leftists. Hopefully the rest of SV/Bay Area companies follow suit.
I just cannot fathom why people would use a tier 2 cloud provider over AWS or GCP. It can't possibly be significantly cheaper. The tooling is non-existent. There is little third party tutorials/documentation compared to tier 1.
Because (comparing DO to AWS) it is much cheaper for many use cases, there is good tooling and documentation, and the user experience is dramatically better. I'm curious - on what did you base your assumptions?
Years of tier 1 experience and trying to replicate those workloads on DO to see what the fuss is about. They are not remotely the same. How is the user experience better? Third party support like Terraform is objectively worse for DO
They lost me after I couldn't switch from CentOS unless I also agreed to lose my IP address. Reached out to support and got the unhelpful reply of "use another OS version within the CentOS family" which didn't really help considering CentOS was charging its direction.
Having functioning and accurate reverse-DNS is required for mail servers, as lots of mailservers reject mail from servers without correct reverse DNS. There's likely other protocols where having functioning reverse DNS is a necessity or strongly advantageous as well.
I thought support could help me out given the CentOS situation and it being a special case but they weren't even aware of what was happening or that users would need to move off despite my explanation in the ticket. Hopefully it's better now it's more widely known.
> runs an OS that is in the same family as the Droplet you're rebuilding
This line in the documentation might be misleading. Unless you have a very old legacy droplet that uses HV-assigned kernels, you should be able to reimage your droplet and use another OS.
So anyone who's offended by content you might host on DO can report your site and some identity politics obsessed bureauocrat decides if you're cancelled or not...? The public cloud has turned into an Orwellian nightmare.
If you are worried about 'offending people', moving away from AWS sounds like a good idea. AWS's 'Acceptable Use Policy' [0] states: You may not use [...] the Services
[...] to [...] display [...] content that is [...] offensive.
However, moving to D.O. is a bad idea, because they have virtually the same policies. They can pull the plug if they want to, because it's their service.
That seems like a reasonable list of stuff I wouldn't want to have my platform be used for, either. Though, one might reasonably take issue with that "Other" category, but note that--at least in the US--web hosts cannot be forced to host any content they don't want to.
I think you’re downplaying the “Other” category, which literally gives DO carte blanche in cancelling those they don’t approve of, or worse who the mob pressures DO to cancel.
Do you not see how problematic this is? I agree with all of the other categories wholeheartedly, but even then, application of the policies of these organizations should be uniform, i.e., if both right wing and left wing websites publishes “Violent Threats and Harrassment”, both should be taken down. That’s not what we’re seeing in practice from the major cloud providers.
“Their platform” has become the public square. If you think allowing unfettered control of such a utility by a small number of mega corporations - with political biases that change to suit their bottom line - is a good thing then I’m not sure we’re on the same page, in the same book, on the same planet.
These companies, and the systems and platforms they own, have not become "the public square" any more than a widely read newspaper, or the only Kinkos in town have. Nor are they utilities.
You're using words and phrases with significant legal implications to characterize things that don't meet the well-understood legal definitions.
I don't think you're being dishonest, but I fear you've been bamboozled by the dishonest arguments of others, and are now parroting them.
I may not like the decisions made by private companies, but private companies doing things I don't like w/r/t hosting speech is preferable to getting state power involved.
I don't like Nazi's marching in Skokie. And I don't like what the right insists of calling "cancel culture". But both of those things are constitutionally protected for good reason, and those rights are worth defending.
Section 230 is woefully naive and written for a bygone era. Censorship of speech by a small group of kleptocrat mega corporations on the greatest communication medium ever known to mankind is antithetical to everything America stands for.
Yes, I have considered that. But, no, that’s not my “problem”; I understand that this is not covered by the constitution. I believe that Section 230 should be updated and the clearly vague “otherwise objectionable” language should be removed in favor of specific language regarding where, when, and why content can be taken down, i.e., only when that content is unlawful.
I understand that some are still basking in electoral victories that were without question aided by the ambiguity in Section 230; this euphoria shouldn’t be confused with righteousness.
Imagine a world where Section 230 doesn't exist. A mean corporation takes down your post, or bans you.
What cause of action do you imagine you'd have that wouldn't be barred by 1A?
By and large, 230 acts merely as a procedural fast-last to dismiss suits earlier than they'd otherwise be. But they'd ultimately be decided the same way on first amendment grounds.
230 needs to be updated, not thrown out; 230 clearly has important aspects that must remain, and other aspects that are clearly being abused by big tech.
Oakmont, Inc. v. Prodigy Services Co. was in my mind and many others greatly preferred to the current mess we see caused by 230.
Please tell me how your 1A hypothetical works out. Genuinely interested in hearing how you think 1A and SCOTUS will support your side of this argument.
Stratton Oakmont v Prodigy Services is only tenuously related to your gripe about websites removing content. Stratton didn't sue Prodigy for taking down posts by Stratton. They sued Prodigy for not taking down allegedly defamatory posts by users.
I asked you to provide a cause of action for "censorship" by a web site. SOvP isn't that. A more recent, federal, and better-fitting case would be PragerU vs. YouTube. There, PragerU sued over their content being remove/restricted. They lost on 230 grounds. But even without 230, they almost certainly would have lost on 1A grounds, because YouTube's exercise of editorial discretion about what content they host is clearly 1A protected activity.
Going back to Stratton Oakmont for a moment: I understand the argument that, absent the liability shield in 230, content moderation of less-than-clearly-illegal content would be vanishingly rare. Web sites would seek to avoid potential liability by getting out of the content-moderation business, which would fix your issue. I'm not convinced that would actually be the outcome. Even if it were, I wouldn't enjoy every web site devolving into 4chan-but-covered-in-spam.
Okay. But when Twitter or whoever takes your post down, how does treating them "as a publisher" help you? It doesn't. They have a 1st Amendment right to do that.
Why wasn’t Stratton Oakmont v Prodigy Services overturned on 1A grounds?
“... if they are neutral platforms, they should have immunity from litigation. If they are publishers making editorial choices, then they should relinquish this valuable exemption. They can’t claim that Section 230 immunity is necessary to protect free speech, while they shape, control, and censor the speech on their platforms. Either the courts or Congress should clarify the matter.”
The practical reason it wasn't overturned is that it wasn't appealed (IIRC, they reached a settlement). Had there been an appeal, it's possible the court would have recognized that exercising some editorial control over third-party content doesn't make you into a newspaper-like publisher. They might have set the bar some place like "actual knowledge of the content, pre-publication". That might have created a precedent where there was something resembling distributor liability - where a platform wouldn't be liable unless they were put on notice and failed to take down defamatory speech.
That would still be a problem (because it would allow companies to censor critics, even if the criticism wasn't actually defamatory), but at least we wouldn't have incoherent arguments about a non-existent publisher/platform dichotomy.
That city-journal piece is pretty awful. For instance, regarding the bit you quoted, they absolutely can claim that 230 protects free speech. Private censorship on private property is free speech.
Congress almost certainly can't pass a law that outlaws viewpoint discrimination for web sites that would survive 1A scrutiny. Fiddling with defamation liability to get around the constitution might work, but just because you can doesn't mean you should.
you do realize that the folks upstream to DigitalOcean, meaning the ISP and data centers, have their own AUP that DigitalOcean is required to meet in order to be their customer?
Every ISP has an abuse process, this isn't very unique here.
I use DO because it's simple and easy. I got tired of dealing with AWS corporate-sized crap, all I want is to spin up several servers using terraform and get work done. They way you can automatically apply firewalling rules to server tags, for example, is exactly what I need.
And don't even mention Azure, where I feel like in a huge maze of mirrors. Nothing is simple, and you waste lots of time on clicking and figuring things out.
So yes, DO has very real value, and I'm glad they exist. The only roughly comparable provider I found was Vultr, which I like quite a bit, too.
Doing anything with those big vendors is an absolute headache, and requires a lot of vendor-specific knowledge/experience, which is expensive (and their offerings are generally expensive and/or harder to estimate pricing on).
DO/Linode and others are perfectly fine serving a sizable audience of users who do not need the complexities that those larger vendors offer.
I struggled for days trying to add more disk to an existing ubuntu image on Azure in 2018. Documentation sucked - both what was there and discoverability. Their instructions just... didn't work, the UI was incredibly .... Azure wasn't my choice - client had O365 and their 'IT vendor' had 'credits' on Azure, so... I had to go with Azure... and it was a huge time sink to do some basic stuff that is literally a few clicks with DO and Linode.
Yes, I don't think I can create private VPNs in the cloud with DO - I don't need to for most projects. The level of functionality DO/Linode/etc provide is adequate for a lot of projects, and they are growing their new functionality to serve needs in a way that seems usable by people without needing to be a certified expert in BigVendorCo.
And... there's nothing inherently horrible about lightsail, but if I've decided that's all I need, I may go with other companies that have that model as their main focus. If I decide I want more of the AWS infrastructure, I'd stay in AWS (or... GCP, or Azure, or whatever).
The company I work for solely uses DO. It has all the things you might want that won't get you cloud vendor locked. Plus their UI is actually useable, unlike my experience with AWS.
Many SV startups just end up taking their VC investment and handing it to AWS over a period of 18 months.
This. DO, imo, crushes AWS's pricing, documentation and ease of use for anyone but big enterprises. I used their managed k8s for a personal project as someone who's never used k8s before and it:
* didn't cost more than the $10 droplet it ran on
* just worked
* super easy to operate
* has had no issues so far
If I were working for any kind of not-ridiculously-large business that wasn't connected to an endless funnel of money (VC/adtech...?) I'd pick DO.
The last time I checked Lightsail uses similar CPU credits[0] as their t2 ec2 instances.
As long as you're only using a tiny portion of your CPU it's fine but if you start doing work on your instance where your CPU is being used for a sustained amount of time then you run out of CPU credits and performance is drastically degraded.
DigitalOcean has no such mechanism. I've never had a droplet's CPU performance get penalized because I used the allocated resources that I was paying for.
From what I've seen of benchmarks online, DO's droplets have better sustained (and often overall) performance than Lightsail, and this looks to be the reason why.
(Which isn't to say people should avoid Lightsail, but I don't think Lightsail is the obvious 1:1 replacement for DO that a couple of people in this thread want it to be. Lightsail intentionally restricts various things to avoid cannibalizing their normal AWS sales.)
Their VPS pricing is competitive but pretty much everything else isn't. Specifically, I'm looking at EKS vs DO managed k8s, block storage for VPS's, and so on.
DO charges $0.10/GB per month for block storage, the same as AWS Lightsail charges for block storage. AWS also has Lightsail Containers which are far cheaper than EKS and is their DO/Linode/etc equivalent.
Lightsail also offers load balancers for $18/mo compared to DO's $10, $30, or $60 per month.
DO $15/mo managed DB 1G 1vCPU 10G SSD
AWS $15/mo managed DB 1G 1vCPU 40G SSD
I'm really not seeing how "pretty much everything else isn't" with respect to their offerings.
Huh, you're right. I guess the main place DO beats AWS is in their managed k8s pricing, which is funnily enough one of the main thing I use. Is lightsail containers more like docker or more like k8s? It reads to me that it's more like docker.
I'd argue that the billing is much more straightforward.
For example, you just pay $5 for a monthly droplet. AWS offers a near competitor to this,AWS light sail.
Digitalocean is easier to use, but only marginally. I strongly suspect many users who find AWS to be a confusing mess try DO instead. I used DO for a long time , but I've since switched to AWS
Do you think Kia or Toyota (the brand, not the company) have a long term play, or is everyone going to shift to Lexus, Mercedes and BMW?
Not everyone is interested in paying the exhorbitant pricing the big cloud providers charge, regardless of features.
Now they might get acquired by the big boys and run under their existing brands as a low cost alternative. But the services aren't going away. I'm willing to bet the big guys want nothing to do with them though. It'll just eat into their margin.
Easy setup and maintenance, and reliable pricing. Order a Unix server for $10 per month, 1TB traffic included, and you get a server with fixed hardware specs. If you get more customers, you scale the server and pay a higher price. On AWS and others: spin up a server that costs more in comparision to Digitalocean, and prepare to find costs on your bill you never thought you would have: traffic, permanent storage, logging,...
I don’t own a business just personal projects, but one reason why I prefer DO over AWS, GCP, and Azure is the billing aspect. With DO I just buy credits and if I go over the limit they cancel my services, I’m fine with that. With AWS I could rack up an insane bill, I’m not fine with that. Notifications about billing is not the same as suspending all services.
This is why I like DO. If other cloud providers do have this please lmk because nothing makes me more nervous than having my personal CC on file with very little retribution as a consumer (I don’t want to get banned from all Google services for doing a charge back).
That is the target, AWS you need a math degree to understand the billing and even then you will never know exactly what you will be billed until you are actually billed...
This is not the case with DO and Linode
Sure for a large enterprise that needs that complexity and flexiblity to squeeze every penny out it would not be a good fit
For many use cases, the included free bandwidth and no "oops I just spent a lot more than I intended to" features make DO vastly superior to AWS/GCP/Azure.
While I wouldn’t use DO et. al. for my production environment (it’s on AWS), I’ve got a handful of ancillary services that don’t need the complexity of AWS.
They’re so small and inconsequential to the business that I don’t want to spend any time thinking about the security risk of keeping them anywhere near production. They don’t need access to prod data so why put them in a position where they could potentially allow access to it?
So, for me, the easiest thing was to be put them with a completely different company. Don’t have to worry about any potential issues with imitating it on AWS (via additional accounts, different region, etc) which are all prone to human failure over time.
Maybe I’ve a bad reason, but it works for me and DO gets ~$50/mo from us and I’ve never considered switching.
Simplicity. As a solo dev I find AWS and GCP to be a nightmare. I do t have time to learn either interface, and not doing so is dangerous as not knowing the inner workings can have security or reliability consequences. Just look at all the exposed S3 buckets over the years.
Just picking a VPS from Amazon is an absolute nightmare. All the different levels, with the credit bullshit.
DO I can log in, pick a VPS with the specs I need and be done with it. I don’t need to worry about over using it and running out of credits and being throttled and having that to troubleshoot.
DO has been reliable for me, affordable, and incredibly easy to understand and use.
I think the value prop is simplicity and customer service. DO/Linode/Vultr offer a way simplified product that's perfect for the long tail of users that don't need all the more advanced aws/gcp features. Plenty of great businesses are already being run on these smaller platforms. And you just have to read one of the GCP account horror stories on here to get the opportunity for better customer service for SMB customers
Personally I'm really proud of the work we did, and I'm overjoyed to have been part of building a business that is going to be a public company. Thanks to the HN community for being so supportive over the years.