Re: MySQL drops support for most distributions

On Wed, 2006-12-13 at 14:21 -0500, Tom Lane wrote:

I think the real criterion for big companies is not so much whether you're supporting your "own" product as whether you're big enough to be worth suing if things go wrong.

This is a common misunderstanding and it is incorrect, at least in my experience. I work at a company with >10K people. I oversee computer architecture and operations for Research (~800 people) and I work very closely with our large IT group.

In order to understand how we purchase hardware, software, or support, you have to understand what's important to us. A successful company must focus on their products and not irrelevant details about how they gets produced and delivered. Employees may personally care about the detailed means to product, but successful companies and their managers -- and, ultimately, customers and stock holders -- do not.

The major concerns for our purchases include: 1) Does it meet our functional requirements? 2) Does it integrate with our existing infrastructure? 3) Can we identify a support channel? and 4) What's the risk relative to other options? These days, OSS packages frequently exceed functional requirements over proprietary alternatives. Apache is an irrefutable example. Big vendors often have proven track records for (2) and (3), but it's not the bigness per se that appeals. We choose small vendors when that's appropriate for a need. Whom we sue when things go wrong is almost never a consideration during purchasing. If a relationship goes south, a suit is unlikely to address our primary goal, the product.

Now, lest you think I'm a corporate troll on the pg lists, I should tell you that I'm probably among the most visible and vocal open source supporters here. I've long railed against proprietary software -- not because of support issues but because I view *some* proprietary software as a real threat to our long-term success. What's important is that our data are usable in ways we see fit, without encumbrance from vendors. This is not the goal of big vendors who require depend on lock-in for growth.

The EnterpriseDB folks have the right strategy. Nobody wants Oracle itself, but rather they want database services that behave like Oracle (er, except the parts that annoy). If I can't tell that I'm not talking to Oracle but getting the "right" answers, why should I care? Cheaper too? Even better. Oracle should be scared because it seems inevitable that their database business will be commoditized out of existence.

Concern for risk is perhaps the most elusive problem for OSS providers and supporters. Companies don't like risk, and *any* change to a working process is a risk. Much to my chagrin, this risk makes it difficult to unseat even mediocre products. We should all cheer EnterpriseDB's success in booking some big name companies. Their successes will establish PostgreSQL as a reliable, cost-effective, and empowering alternative to proprietary databases and therefore decrease the risk concerns.

The only reason I spent this much time weighing in is because I'm thrilled with PostgreSQL (er, sorry Tom, Postgres) and appreciate and respect the terrific work done in this community. Thank you.

Cheers,
Reece

-- Reece Hart, http://harts.net/reece/, GPG:0x25EC91A0 ./universe -G 6.672e-11 -e 1.602e-19 -protonmass 1.673e-27 -uspres bush kernel warning: universe consuming too many resources. Killing. universe killed due to catastrophic leadership. Try -uspres carter.