Postgresql is actually good enough to use in enterprise applications. I've even seen MySQL used in a large retail chain. When I see Oracle used, it's for one of two reasons: compatibility (lots Forms and PL/SQL code) or a manager playing safe.
Maybe we should compare software with other types of goods, such as capital or luxury goods.
Postgresql is actually good enough to use in enterprise applications
Anecdotally, Postgres is running some "you can't possibly be serious!?!" applications at some of the largest Japanese companies. The big one we would point to get to clients to switch from a Certain Enterprise Database is NTT ("the phone company"). They've really adopted it with a vengeance and are sinking eyepopping amounts of R&D into it.
Databases are not really my thing, and after you get past "bigger than a payroll system" my ability to comprehend matters of scale drops off fast, but I nearly hit one of our guests with a face full of tea when he told me a particular app which they have on postgres.
A) Running a payroll system is not a difficult software problem. A 400$ pc is significantly more powerful than the 80's mainframes that ran huge enterprises. The only real issue with such systems is the terrible state of in house “enterprise” software.
B) However, for those rare cases where you really need a monster of a database there are plenty of options. Teradata for example can be used when hundreds of HDD are just not enough.
So while Oracle can scale up (to B), it's rarely used in that capacity simply because it's not needed (see A). IMO, the only need to use Oracle is when it's possible to scale rapidly and money is not an issue.
> Funny how the most upvoted comment here has pretty much nothing to do with the premise of the article. This community is going down hill rapidly...
Firstly, the comment you refer to IS relevant because the article explicitly states that PosgreSQL cannot compete with Oracle, et al in the enterprise. I mean it's right there in the conclusion. Maybe it needs more info to back it up, but then so does the article itself when it makes its point.
Secondly, the reason I would wager on it being the highest rated comment is because the greatest value of HN comments is when you finish reading one of these articles and are quite taken in by the premise, many times (for me personally) right there at the top of the comment page is someone with a different experience reminding you not be so impressionable. In my time here this quality of HN comments has been consistent. In this case I would say patio11's fleshed out response is more useful (edit: and already it has overtaken the parent in votes - I would hold off any emergency quality control measures for the time being) but the cream usually rises to the top.
> I wonder if this type of degradation is inevitable when the masses invade
When I read a reaction like this to someone's comment, I always find it interesting to click on the user profiles and see when both parties joined. In every case I've done it, the person making a comment about "the masses" invading has been here a fraction of the time of the person they're referring to. In this case, your 111 days is barely more than a quarter of the time forinti has been registered.
For a long time, Sun has believed that their business interests (selling hardware) are served by making software free (or as cheap as possible).
Being aware of that "business interest" provides a lot of context to this post (which seems to criticize software vendors for the fact that their products aren't free or that their product-upgrades may not be free).
While Sun's business interest often masquerades as love for the consumer or for the open-source ideology, this post does deserve some credit for an acknowledgment that free services generally means a business model where you charge consumers for support, services, "other software", hardware etc.
> Software companies are required by SFAS No. 86 to capitalize certain development costs of software to be sold, leased or otherwise marketed. Capitalization occurs once technological feasibility has been reached and costs are determined to be recoverable. Capitalization ends and amortization begins when the product is available for general release to customers
Support costs vary with respect to number of customers, and so are a true, and often overlooked, variable cost for shrinkwrap software companies. While many companies generate revenue with support, the basic level of support that customers reasonably expect (e.g., a web forum where people will make some effort to address bugs) is non-trivial to provide.
VC = TC-FC FC is always in the long run equal to zero, so VC = TC Marginal cost Which is what is seems like you are talking about is MC= d VC/d q where q equals quantity output. So in this case q would be the number of copies of the software produced or sold. Not necessarily the number of customers.
He fails to mention that in the long run, ALL COSTS ARE VARIABLE! So his whole premise is just an explanation of a very brief moment in time. Also software does wear out. As Technology increases it eventually becomes obsolete.
I would argue that giving away software does involve a cost. There's the physical cost (maintaining downloads), but the larger cost is foregone revenue, an opportunity cost. It's a theoretical point, but worth mentioning.
I wonder how soon the O from FYO will take this post down? For those who haven't noticed, this post was written in 2004. While I was reading it I thought this must be the guy's last day at this job.
Maybe we should compare software with other types of goods, such as capital or luxury goods.