What's the deal with SQL Server 2008 implementation?

I was talking with a Microsoft representative at PASS and he asked me about our SQL Server 2008 implementation plan.

"I don't have one."

"Why not?" astonished look.

"It's up to my ISV's to certify their applications, and they're not doing it.  I have software on SQL Server 2000 still, the ISV's aren't upgrading it and won't support it if I upgrade."

Another astonished look.

"So what can we do to fix that?" he asked.

"I don't know."

I didn't know then, I have a few ideas now.

The problem is that the people who write software that work with SQL Server and then sell that software (I'm not talking about things like SQLBackup or Hypebac.  I'm talking about ERP systems, line-of-business systems, and so on) don't have incentive to certify that their applications work against the new SQL Server versions, that they don't have the QA resources to do the certification, and they don't make any money providing us with an upgrade because it's not a major release for them.

That's a problem.  That's why I still have about a half dozen SQL Server 2000 boxes, and exactly 0 Sql Server 2008 implementations.

What's the solution?

Automated QA.  Which has its own set of problems, the chief among which is that the licensing is EXTREMELY expensive.  Like "Go buy yourself a nice car.  Italian car.  The kind that gets to kick off Top Gear every week."  That expensive.

And that's outside the reach of ISV's.  They drive nice Hondas, not supercars.

So, how do we fix that?  Well, the Automated QA vendors I've worked with in the past have had a business model of selling a few really expensive licenses to very large companies.  This keeps the marketing costs low (smaller number of CIO's to take to golf outings), the support costs low (when you only sell a few licenses it limits the number of simultaneous calls you can get) and keeps the software out of the hands of a lot of people who really need it.

This is a situation Microsoft has been in (and been victorious in) before.  IBM had the same business model with their mainframes.  Oracle had (still has) the same model for selling thier DBMS.  In nearly every market with this business model in the personal computing space, Microsoft has built a product, sold it at a reasonable cost, and forced the existing market leaders to rethink their strategy or perish.  And made italian supercar level money by selling to hundreds of customers instead of a dozen.

So, if Microsoft wants to increase SQL Server 2008 licenses, they need to enter the automated software QA space.  Then start applying pressure to ISV's to adopt automated testing procedures with Microsoft software, because it'll reduce their costs without breaking the bank.  The pressure doesn't need to be negative pressure, more of "hey, here, have a huge discount on this software if you'll implement it," or "This is part of your Partner-level benefits."  Build a great product, support it, and sell it.  The big vendors in this space can already show a good ROI, if the initial cost of entry was drastically reduced then the ROI would be even bigger.

The ISV's build testing scripts for their applications so they can test all their minor changes.  Then along comes a new version of SQL Server, and the certification process is "run the same script we do for a bug fix, just point it at a SQL Server 2008 box".   It certifies, or it has a few little glitches that get cleaned up, and the product ships.  Way cheaper and faster.

Then the ISV's are happy because they've got better software from having better processes, Microsoft is happy because the ISV's are making better stuff, they get to sell more boxes of software, and people will start upgrading their back office components, and finally the folks like me are happy because we get to consolidate some versions together more quickly.

Faster please.