Saw the Fortune article and passed it on to my client contact ;-).\r\nI've already had this conversation twice this week.

I agree with much of what Barry wrote, regarding the dynamics and\r\ndifferences between OSs and databases. Existing large data systems\r\nprobably won't change for years, if not decades. Hell, there are still\r\nVSAM and CICS systems running on MVS (don't ask me what these are, I\r\njust see the contract ads for 'em). Legacy can hold you glued to a spot\r\nfor a long time. I do expect the trend to be clear\r\nwithin three years, though the endgame may not play for some time after\r\nthis -- much in the same way that GNU/Linux was clearly long-term viable\r\nin the broad market in 1999, though widespread server adoption was still\r\na year or two off, and desktop deployment is still very minor.

About two years ago I suggested that after the OS and desktop apps\r\nspace, databases would likely be the next major free software vs.\r\nproprietary battleground. The dynamics would be different because of\r\nthe many lock-in factors that Barry mentioned. This changes the\r\ntimeline, but not the ultimate outcome.

Confound the issue with the fact that low cost versions of\r\nthe major databases are available for either limited implementations or\r\npersonal/professional development. I've ranted here recently about\r\nSAS's seed-grain poisoning pricing. Oracle has long followed a practice\r\nof making developer versions available at low cost (generally there's a\r\n"personal edition" bundled with various Oracle texts (e.g.: the\r\n"Unleashed" series, and similar). So I don't expect developer cost to\r\nbe as much a driver as deployment cost and, ultimately, features.

Existing customers of proprietary databases, similarly to\r\nexisting customers of high-end computing environments, will be the last\r\nto convert. Of course, that's classic Christensenian bottom-up\r\nencroachment. This is the typical innovators' dilemma cycle.

At the same time, smaller firms, new ventures, and technical\r\ncompanies are starting out, or switching, to free software databases.\r\nGoogle, Yahoo, and Slashdot are mentioned prominently in the Fortune\r\nstory. Less well known is that many dot-com sites went up on MySQL and\r\nPostgres, but went down on Oracle (the Emerald City couldn't stand the\r\nfact that some of the biggest, fastest sites on the 'Net were\r\nnot running Oracle). I know that the eToys conversion was a\r\nnightmare, and unwarranted from a technical standpoint, according to the\r\ncompany's former technologist. Again, adoption at the low-end of the\r\nmarket is classic Christensen. This is an expansion of the\r\nbase of sites that can use full-powered databases.

In the long run, proprietary databases will either be supersceded, or\r\nconverted to free software. Both trends have been in evidence: MySQL\r\nand Postgres are pure-play free software databases which have replaced\r\nor been used in the place of proprietary DBMSs. Borland and SAP show\r\nthe other tack -- products which have been non-competitive in a\r\nproprietary market were open sourced. Many see this as a last-ditch\r\nend-of-life play, but then we saw similar actions from the likes of\r\nCorel and StarOffice, that bloated stuck pig of an office suite, the\r\ndesktop OS and apps space. Corel's dropped off the map, but we now have\r\nOpenOffice which is pure free software. The long-term viability of such\r\ntransitioned products is a significant concern.

One significant aspect of the database market is that there aren't\r\nmany players. Of the old "big four", Informix and Sybase have largely\r\nfallen (Informix being bought by IBM, Sybase was the basis of MS SQL\r\nServer, though it remains a freestanding product). Oracle is the only\r\nmajor pure-play DBMS company. IBM's got its DB2. Then there's\r\nMicrosoft and MS SQL Server.

Oracle's got the most exposure to bottom-line impacts from\r\nencroachment. The first signs will be a slowing of market growth, then\r\nan attempt to maintain revenue growth by squeezing existing sites harder\r\n(the same lock-in that prevents mobility will make this an effective\r\nstrategy). I'm seeing similar trends with SAS, who've seen their\r\nmainframe revenue contribution increase from 33% in 1995 to 48% in\r\n2000.

The market structure could affect strategy. Oracle is #1 in\r\ndatabases, IBM #2. Combined with IBM's involvement in free software,\r\nthis could accelerate trends. IBM could attempt to cannibalize Oracle's\r\nmarketshare by opening up DB2. On the other hand, Oracle already has a\r\nsignificant services and consulting division. Oracle could\r\nfeasibly survive by switching to a greater emphasis on these revenues\r\n(preferably with some form of a subscription model, though to what is a\r\ngood question, to even out revenue flows which tend to be highly\r\nvolatile in consulting). If they were to go this route, expect to see a\r\ngradual trending toward free software licensing, though likely not\r\noccurring in a single step. Oracle, like all proprietary software\r\nvendors, likely has a great deal of encumbered code which would be\r\ndifficult to liberate, prolonging the effort. The code is also likely a\r\ntangled mess. Both of these issues have plagued OpenOffice -- portions\r\nof the original StarOFfice code had to be tossed, and the One Big\r\nExecutable model of SO is the reason for it's being a bloated stuck pig,\r\nas I allude above.

The one clear loser is Microsoft. Free software is untenable with\r\nthe company's existing business model, and likely with its existence.\r\nWhile there's a survival path for both Oracle and DB2, MS SQL Server is\r\na doomed product. They can't open their source (actually, a strategic\r\npurchase of Sybase by a significant free software player would allow for\r\na SCO play down the road...Hmm...), and they can't compete against a\r\nzero-cost product, or the development capabilities of free software.

Edit: "In the long run..." tidied up.

...what the heck I was trying to say. Hmm. I'll clean that up a\r\ntad.

Thanks. I hate it when I write sloppy.