A few thoughts on the Sun MySQL backtracking

OK, now that all parts of MySQL is going to be free, there are a couple of things which still stand out.

If you want to do a tiered service of free, Incident support contract, service contract, and variations there of, just don't do it immediately after an Open Source buy. This seems pretty obvious....in hindsight.

And when you backtrack, you need to generate the same level of notoriety/publicity that caused you to backtrack, so that people know you came to your senses. This is even more difficult in this Internet drive-by visit,10 sec attention span.

This was an Open Source buy, which means proprietary products selling tactics need to be done differently. If the selling had been along the lines of ; upto 4 CPUs no charge, anything higher, support will cost you, it would be a big deal with respect to OTHER Database vendors.

One, the vast majority of LAMP servers are on low end stuff and individuals tinkering with it, so no one will probably complain on the pricing front.

Second, you raise the bar for the other database vendors by cutting out their low end revenue segment (Oracle editions are free on single CPU machines and for personal use). Forced to compete with MySQL on multiprocessor boxes, you drive the database costs lower, cannibalizing prices in the process.

Three, if you benchmark it on your hardware and make it come out on top,(it's your product after all), you make a good TCO case with zero software acquisition cost for the customer but backed by your company support on YOUR hardware. You use the software to drive the hardware sales.

Heck, if Sun's multicore is as good as they claim, benchmarking on lower Mhz chips with more cores should show some results to crow about?

Let's see how this pans out.

Proved Right!

Back then, I speculated that the reason Sun bought out MySQL was probably due to its spat with Oracle over pricing multicore chips.

Well, this story sort of confirms that.

Pricing on a processor based system is now more complex with cores more than 1. and it's going to sock the first guy with more cores per chip, from their trusted partners.

The coming Open Source Database wars

Well, it's the commercial vendors who will be duking it, rather. With the recent Sun acquisition of MySQL and IBM throwing some dough to PostresSQL, (Enterprise DB to be pedantic), things are getting interesting.

I think it has less to do with database feature sets and more to do with pricing. If I understand something about database licensing, it's done mostly on a per CPU basis or number of cores.

Pricing will determine what kind of technology adoption will happen on multicore chips. While increased performance may be got by putting 4,8 or more cores on the chip, the pricing done on a per core basis will undo all the technical advantages.

If two vendors have the same number of chips on similar systems but one has more cores than the other per chip and the pricing for the Oracle database is done on a per core basis.....you can see where this is heading.

Disaster, for all the technical progress made.

I think that's why SUN bought out MySQL; From Oracle's point of view, putting more cores on a single chip means less revenue as less chips/system does more now. The only way it can make money is to switch to a core based licensing revenue stream. That would mean, SUN's systems would come out costlier compared to a similar CPU based system from IBM, say.

SUN and Oracle has been close for years and the sudden acquisition must mean that the low end SMB market is going to get poached from Oracle.

Soon, expect new TPC benchmarks from SUN using MySQL as the database.

Then the message would be loud and clear to Oracle. As a matter of prediction, Oracle would cozy up to IBM and optimise things for their Power chips I guess.

Update: see this link for more confirmation.