Link doesn't work for me. From another source I found:
Free – Experiment in your own dev or demo app with a web and a worker dyno for free. Sleeps after 1 hr of inactivity. Active up to 12 hours a day. No custom domains. 512 MB RAM.
Hobby – Run a small app 24×7 with the Heroku developer experience for $7/dyno/mo. Custom domains. Run a maximum of one dyno per Procfile entry. 512 MB RAM.
Standard 1X, 2X: Build production apps of any size or complexity. Run multiple dynos per Procfile entry to scale out. App metrics, faster builds and preboot. All Hobby features. 512MB or 1GB RAM. $25 or $50/dyno/mo.
Performance – Isolated dynos for your large-scale, high-performance apps. All Standard features. Compose your app with performance and standard dynos. 6GB RAM. $500/dyno/mo.
Domain redirects are useful for portfolio apps if you'd like to have some coherency in your urls, and maybe you're not trying to showcase that your app is on Heroku's free tier at that moment. That's legitimate, it's also a use case where 12 hours/day is likely sufficient. However users breaking TOS and keeping those apps awake 24/7 so recruiters don't have to wait ~30s for them to load is probably a problem Heroku's trying to solve.
well amazon's free tier allows you to have a smallish machine up 24/7 (for a year), seems to me heroku is going to push away a lot of smallish side projects that might have evolved into actual usage.
Is there anyone outside of a total beginner or a newly created company who hasn't had an AWS account for over a year now? I don't even use mine a lot (mostly S3) but signed up very early on just to check it out, as with most things, so I wonder how relevant Amazon's free tier even is now.
you might have an AWS account but not necessarily to run a web app, cause you're in the "meh, I'd rather do git push heroku now than learn to setup elastic beanstalk/opsworks+docker/ec2+deis".
The latter becomes more attractive now, and once you have learned to use AWS's X tech to run $SILLY_SERVICE you might end up using it for $SERIOUS_SERVICE.
I am in charge of a high volume, medium sized deis setup, and have been actively using deis since 0.2.0, but I don't have that kind of setup for my personal blog. It's overkill. Heroku is easier and allows me to get one with development, so I use Heroku.
It's not so much a "meh, I'd rather do git push heroku now than learn to setup elastic beanstalk/opsworks+docker/ec2+deis" than it is a "I just want to develop a quick idea without wasting time on ops".
So, it seems that the change for production apps is that you can now somehow commit to a monthly price for dynos and get a discount.
1X – $25/month compared to $36.5/month (by the hour).
2X – $50/month compared to $73/month (by the hour).
PX – $500/month compared to $584/month (by the hour).
I tried to post this yesterday[0] and fell off the new stories list, but such is HN.
Basically, they're nerfing their free tier by charging $7/month for a 24/7 running app. People wanting to run a free app may look at App Engine or AWS's one year free tier, or compare the price among other "cloud" and cloud-ish services.
While it doesn't seem certain any price change has happened. A drop in pricing has been way overdue, the cost of computing power and memory has been dropping but not Heroku pricing. I get that Heroku is meant to be a "premium" cloud product, but it often doesn't feel like one.
This also may not be much of a drop for people with minimal production requirements. Under the free hours system, 2x1X was $35 now it may be $50 unless the Hobby pricing is factored in for the first 1X and it's $25 + $7 = $32. For 2x2X the price drop will only be $6.50. Still it's good that the cost to scale has gone down.
The paid hobby tier is actually welcome change, the limitations of the free tier (often used for staging and demo environments) were (are?) annoying. The ability to run 1 dyno per Procfile entry (instead of per app) is a welcome change, as allows you to fully mimic production without dyno scaling more easily at a lower cost. People were likely using while allowing them to pay a reasonable price for these low-traffic systems.
I have been saying for a while 1X dyno is underpowered and unnecessarily memory limited considering it's price (particularly for Rails apps) even before 2X was announced. If I had my way the 1X would be dropped in favour of the 2X at the same price and the 1X only used for hobby and things like a worker role handling infrequent job queue processing. These price changes come pretty close to doing that, while adding a paid, but low-cost, hobby tier for applications with very limited needs.
GAE lost a lot of users in their price hike a few years back, so on the mid/higher end it makes sense. But, hiking on the low end may "save some headaches" but it puts off potentially moneyed ventures "with the bathwater" in a "no public restrooms" way. (Such "signs" are mostly pointless and just make the "shop" look discriminatory.)
I run my companies hubot on a $5 a month Digital Ocean instance. It's also far easier than using Heroku because it's just a regular linux server and not a managed/restricted instance.
I'm in the same boat with my blog. I'm not paying $7 for a single dyno when it's footprint is no-where near 512 mb. They should offer a smaller dynos at say 128 mb for $1.75, that I might be interested in.
I'll be interested to see what they do to my existing apps.
Viable business models for B2B businesses that involve charging the price of a fountain soda per month for anything are few and far between. Just charging your card that dollar costs a quarter or two, and sets up expectations that free tiers don't come with. So you may want a $1.75 fully managed PaaS, but it's doubtful anyone wants to sell that to you.
Indeed, which is why they'll probably lose a lot of people.
I'd be happy to put credit onto the account to use the smaller tier, or PayPal it or something, but I'm not paying for a big footprint that I'm not using.
If Heroku actually does make the free tier unusable (12 hours a day availability), and if custom domains aren't free anymore... then Heroku, time to part ways, and no more referrals to others.
This is the reason I chose Heroku as my platform in the first place over competitors, and if my experience is common, it worked out well for them because I have paid for premium tiers on certain apps along the way, and spoke glowingly of them to others.
I'm also still a little bitter that I incurred financial loss on one app because one of their quiet upgrades bonked Ruby 2.1.1 support, causing a big delay in a critical production push. (But I'm more annoyed that this will be about $100/month cost for me to keep my lighter traffic personal apps on Heroku).
Most likely they will argue that you can use it 24 hours a day, just that they will cut the 750 free hours in half.
Does that mean you'll pay $7 a month or $3.50 a month for a 24/7 dyno?
I agree with you though, they might think they're kicking out X freeloaders, but how many of those freeloaders have production apps and how many of them are in decision making positions?
Are you going to move your very significant app to another platform because of this? Seems like an absurd thing to do to save yourself maybe a couple hundred bucks a year...
No, I'm very happy with Heroku. Just wanted to highlight my progression. Freeloader to paying customer. I wouldn't be a paying Heroku customer without a generous free tier to experiment with.
tl;dr : Current paying customers will see a discount of up to 30%. Current freeloaders will have to pay $84/year for functionality equivalent to what they get now.
Anyone who is technical enough to use Heroku effectively can charge AT LEAST $84/hour for their time, meaning that they'd need to move to a new solution in less than an hour to save money. All the hand wringing is very silly.
> Anyone who is technical enough to use Heroku effectively can charge AT LEAST $84/hour for their time
This is very off base in my experience. I know countless beginner devs who swear by Heroku, because it's designed to be that simple.
You're not a 'freeloader' if you have lots of apps on the free tier but production grade ones where you pay premium. You're a paying customer in that case, and that's probably a common pattern with these 'freeloaders' - lots of free apps, but key paying ones.
All in all, I'm not going to pay $84/year for each of my 12-14 personal apps that have light traffic (couple hits per hour maybe). I'm going to pick a new platform and make it part of my workflow, and for the sake of simplicity, I will be moving all of my apps to it, paid ones included.
I have multiple smallish projects that don't need to be up 24/7, but that may evolve into something more. I'm not going to pay $84/year per project to keep them on Heroku. So, I'll end up moving them, and as they evolve, they'll evolve on the other platform.
Exactly. We're not going to pay for something that might never take off, but we if they do, it'd be nice to be able to keep them where they are. Isn't this how Heroku got to where it is?
Seems everyone grows so big that they turn their backs on where they came from. They're no longer Heroku from the block.
Thanks for writing this analysis. I think the pricing changes are long overdue and great news for developers.
I hope Heroku gets to focus more on the people who understand how many hours/year it takes to run a server safely and securely. At the same time, the free tier still exists and 12hrs/day is plenty for code schools and various tinkering.
Sure, no money has to change hands, BUT if you decided to take the hours that you would spend moving projects away from Heroku, and instead do something productive that creates value for a client you could be spending the same amount of effort, and be making more money than you'll save by switching.
That's an excellent point. The $84/year tier also allows users to keep an app always on without needing a second dyno, which used to be the requirement to prevent apps from sleeping.
So, as a fellow hobbyist with 12 free dynos running on Heroku with custom domains, I clearly don't want to pay $84 a month for apps that aren't meant to generate revenue.
It is a sad day.
Where are you moving your side projects?
EDIT: I'm asking since I have no idea where I should go next. Heroku was a blessing.
I'm not sure what you're doing exactly but I'm a member of hcoop and highly recommend joining. It kind of depends on what your use case is though. I'm not vary familiar with heroku so not sure what a dyno is.
Note, these are not the final prices. Its only part of private beta or AB testing of possible prices.
Also the cost of your existing dynos may not change if they grandfather and not change running dynos' prices. Who knows until they actually announce any price changes.
That plus Dokku could make for a nice Heroku clone/replacement. Was a bit buggy when I tried it a year ago, but the main dev has been sponsored by Deis to support the project for about a day a week now, so I'm guessing it's still going strong.
I understand why they're doing this - companies have to make money - I get that.
I'm still annoyed with Heroku though. Cachet (https://cachethq.io) supports the "Deploy to Heroku" button. My experience is that their support is terrible. I've contacted them several times via; Twitter, email, GitHub and had no support from them in regards to this bloody button.
The reason I support Heroku in the first place is because of the free tier. It makes deploying your own instance of Cachet much easier. However, we've been having issues supporting it with Laravel 5's environment configuration anyway, so this is the last drop in the bucket before I pull native Heroku support.
I totally support Heroku's right to charge whatever they want, of course. They're expensive, but they provide good value for the money.
Historically, I have a bunch of personal free-tier apps which sleep most of the time, and a couple which are up most of the time. My consulting clients, on the other hand, have paid Heroku thousands of dollars per month at various points. I also maintain a buildpack.
But at this point, it's time I get off my backside, set up a docker host on EC2, and containerize the stuff I care about. I can probably pack everything onto a pretty small instance, and I've been meaning to deploy some non-HTTP services. Besides, it's closer to what I'm using in production.
Thank you, Heroku, for some very enjoyable years, for top-notch paid service, and for the free hosting!
I think this is great! I was never very comfortable with the free tier where apps are swapped out (leading to a long loading request time) after an idle period.
A low cost $7/month plan for a dyno to always be running sounds good.
You can get around that by using a service like Uptime Robot to ping your site every hour and keep the dyno running. That way you have a free 24/7 app that doesn't go idle. I think Heroku is trying to get rid of that with these changes.
It seems they're still offering a free tier for an app that doesn't need to always be running. But if you need it running more than half the time, you need to pay at least $7/month. Will this affect current customers?
I understand there were probably too many apps running on free tiers, and probably some that abuses the TOS as well. But with AWS and others having cut prices so much, shouldn't Heroku also offer some kind of better discount?
I'm kinda surprised that we haven't seen a price decrease, since AWS keeps reducing prices. I can understand they wanting to decrease the 'freeloaders', but it seems shortsighted, those people are their future paying customers.
Their new pricing makes it more economical to just get a small droplet from digitalocean, i'm surprised they chose to be above that point.
I wouldn't be so quick to assume they're being shortsighted. I have no idea what the details are, but I've got to assume they have very explicit data about who and how many people convert from free to paid tiers and that they're taking this data into account.
Again, assumptions here, but maybe they found that only a very small % of users start with a completely free 24/7 app and move to paid? Maybe the folks who convert start with a low-utilization (<= 12 hrs/day), hence the limit.
I dunno. I'm just guessing that there's some logic behind the decisions.
As a paying customer I am glad that Heroku is cutting down on the free usage. They are using AWS so free usage means actual expenses. Sure, it can be seen as a marketing expense but Heroku is well known now and don't really need it to gain awareness.
I am happy to pay for my production, but I feel as though part of the benefit is that I get a cost-free development playground. If they take away that, I'd rather put the money into something else where I get even more control.
I guess they want to go leaner, which is fair enough, but they will lose paying customers as well.
For a cost-free development playground, they have the hobby tier. You can run your dev apps for 350 hours per month - that's a lot of dev testing time.
Does anyone know when this new pricing starts? I have a free app on here that is only heavily used during certain parts of the year and I am about to hit that busy time and hope that this new pricing won't take effect too soon.
This actually works out well for me since I don't host any "free" projects there, but I suspect DigitalOcean will be deploying a whole ton of VPSes in the next month with the Dokku image running ;-)
IBM is trying to build their hosting business. They offer a really generous free monthly tier on BlueMix. You can run a web app with a small data store for free.
Actually, BlueMix gives you 375 hours of 1 gigabyte units. If your web app is configured to use 0.5 gigs of RAM, then you get 2 * 375 hours free per month. I was running a node web app with a data store using a combined 0.25 gig units, so running that 24/7 would only use up half of the free monthly tier.
I wonder if they will also change the pricing of the database tiers. Since paying 25$ for an 1x dyno and paying 50$ for a database just seems to be out of wack.
The database pricing looks pretty much like before. What is problematic about paying $50 for a database that you don't have to manage, has continous backups, dataclips, and ability to easily scale and set up replication for?
How much would a db admin charge you per hour to do the same?
Finding someone who can setup Postgres wal archiving and replication in only 6 hours per year is surely costing you 600$ in additional salary per year. I don't think your argument holds any merit at the level of hosting costs we talk about. Also as someone who does host stuff on a dedicated machine too, have you looked at what is involved in upgrading to say Postgres 9.3 from 9.1. If you have single server you will just not do it unless you are also big into Russian roulette.
Finding someone who can setup Postgres wal archiving and replication
Your $50 database doesn't need that. A simple backup works fine for a great many people.
I don't think your argument holds any merit at the level of hosting costs we talk about.
In order to match the performance of a measly $50/mo dedicated server you'll pay in the ballpark of $1300/mo to Heroku (16GB DB + 16 "dynos"). In between these two figures there's a lot of merit to be had.
have you looked at what is involved in upgrading to say Postgres 9.3 from 9.1
Yes, you stop the db, run pg_upgrade, and start the db.
But as you correctly state, most people in this bracket have no reason to upgrade their database software ever.
Free – Experiment in your own dev or demo app with a web and a worker dyno for free. Sleeps after 1 hr of inactivity. Active up to 12 hours a day. No custom domains. 512 MB RAM.
Hobby – Run a small app 24×7 with the Heroku developer experience for $7/dyno/mo. Custom domains. Run a maximum of one dyno per Procfile entry. 512 MB RAM.
Standard 1X, 2X: Build production apps of any size or complexity. Run multiple dynos per Procfile entry to scale out. App metrics, faster builds and preboot. All Hobby features. 512MB or 1GB RAM. $25 or $50/dyno/mo.
Performance – Isolated dynos for your large-scale, high-performance apps. All Standard features. Compose your app with performance and standard dynos. 6GB RAM. $500/dyno/mo.
H/T http://notes.ericjiang.com/posts/881