% fortune -ae paul murphy

The storage wars

Over the last year or so we've seen some opening skirmishes in the coming storage wars: Netapps's lawsuit against Sun over ZFS and the consequent counter-charges, enhanced storage announcements from both EMC and IBM, Sun's decision to buy out MySQL's founders, and the recent publication of work Sun's been doing on integrating flash memory in the Solaris/ZFS performance equation.

This week, at Oracle OpenWorld, we saw the beginnings of round two: HP's presence at the head table with Larry Ellison and Compaq's "Exadata" imitation of Apple's X-Serve trumpeted as the next word in storage; Sun almost instantly retaliating with a Jonathon Schwartz blog on saving millions by using ganged thumpers in large scale data warehousing; and, EMC/Dell chiming in with a me too press release on flash enabled storage.

Meanwhile, in one of those backroom announcements most of the media types attending seem to miss, Oracle quietly reconfirmed that the fusion applications they're putting together from the Edwards, Seibel, PeopleSoft, and Oracle applications portfolios will drop both pillarization (breaking large databases into multiple schemas for independent implementation on PC scale gear) and PeopleWorks in favor of a single, fully integrated, database and programming framework running on an OS independent grid made up of small computers.

Oracle 10g, of course, has run on small grid machines for years now, but whether they can make Fusion's process integrated business inteligence, third party interfacing, and controls work that way is an open question - and will, I'm betting, either be responsible for at least another six months of delay in getting Fusion out the door or produce a two step commercial release process in which the grid product follows the SMP release by at least that long.

What's behind all this, however, is what I think of as next generation storage wars: in which Sun pushes order of magnitude cost-performance gains from an open source database environment running on fast SMP machines with flash enhanced "thumper" style storage into easy markets like data warehousing, and everybody else gets forced into indefensible defensive positions.

I'm going to speculate about Oracle's survival strategy on this tomorrow, but just mention right now that the real bottom line on the announcements at OpenWorld is that two other big players in this market, IBM and Microsoft, so far seem to be relying mostly on customer inertia and pretty press releases - betting, in other words, that their IT customers and colleagues will cheerfully continue fooling their actual employers on the value of alternatives.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.