I began to feel like a broken record. Just about every time over the past three years that I spent some quality time with the Sun product planners and marketers, I'd chime in with, "What about the data?"
All my research indicated that long-term infrastructure decisions on business applications and data -- not OS or runtime -- were leading the software strategies of enterprises as they sought to 1) consolidate applications, 2) reduce datacenter costs and complexity 3) service-enable those applications and data, and 4) move toward Web services and/or a services oriented architecture (SOA).
IBM, Oracle, and Microsoft were having a field day in the field talking the consolidated datacenter talk, while ushering the users away from Solaris to Linux or Windows on far cheaper x86 hardware. The consolidated database plus cheaper hardware duet was a no-brainer. Still is. It was plain as the nose on my face.
The role of database in the equation was by no means trivial, and in many shops where they were dozens of siloed apps, the data issue trumped the decision even over business applications (if there was room to wiggle from SAP's iron grip). Oracle was on this like short-sellers on General Motors. Sun needed to go more on their own or with new partners toward apps, or data, or both.
The Sun folks -- as they had for some time with first, Web services, then SOA, then ESBs -- sort of blinked and said databases were not so important. This was while Oracle was dope-slapping them in their own accounts with all the talk about unbreakable Linux.
It was going on three years ago when Sun came out with the Java Enterprise System (JES) as an umbrella under which they grouped and integrated their middleware and runtime. At the time I took umbrage with the name "Java" Enterprise System, though I liked the subscription pricing theory.
I though the "J"ES naming was a huge about-face on the neutrality and supposed church-not-state stewardship of Java as an industry standard. They shifted on my suggestion a wee bit, calling it on some press releases "Sun Java Enterprise System," but the damage was done. It drove the final wedge between Sun and the other major Java licensees, just when those folks had more leverage -- largely based on database and cheap hardware strategies -- in the field than Sun.
Kick 'em when you're down, seemed to be Sun's mantra. At the same time I told them that Larry has unbreakable Linux but you have MySQL, or other open source projects to advance the playing field a little bit toward level on your end. SAP has R3, but you could have open spource apps. Cozy up to SAP, I said. Silence.
Well, Sun finally seems to be finding what I consider a far better open source strategy than their hard-to-figure "opening" of their various market-sluggish commercial products to the CDDL license. (You do notice that market winning products like the directory, ID management, and provisioning and remain non-CDDL.)
That better strategy is to keep their strong products fully commercial, non murky in their license standing, and build out a best-of-breed services delivery platform par none built on integrated and supported combinations of Sun's best (namely Solaris), and the best of open source components, such as database, ESB, business applications, and yes, Java runtime. Make both NetBeans and Eclipse applicable. Aim it all as carriers and service providers first, and enterprises second.
With this week's moves to a more 64-bit (Where is the 33d bit!) ZFS open file system, the snuggle to PostgreSQL, and the momentum of Solaris 10 on AMD 64, things are looking much brighter. Too bad it didn't happen two years ago and the business fundamentals might look brighter too.