The Xserve, which Apple officially put to pasture this week, proved to be another example of the company’s on-again-but-mostly-off commitment to large scale, enterprise-class computing. It’s a market where the company’s track record has been a series of disappointments, but for a while at least, Xserve seemed to have changed Apple’s luck.
Depending on how you do the math, this is Apple’s third or fourth failed attempt at establishing a server platform for enterprise IT. The first failure, the Macintosh Office of the mid-1980s, managed to give us the LaserWriter and AppleTalk, but never delivered the file server hardware that was the heart of that era’s networking implementations. The Apple Workgroup Servers of the mid-to-late 1990s don’t really figure in my reckoning, because the hardware in these was a Quadra, Centris or Power Mac workstation running MacOS with a few server-like add-ons and larger hard drives; they were a kludge at best, and changing the name of the later models to “Macintosh Server” didn’t cover up the lingering stench of lowered expectations. The AIX-based Apple Network Server was an ambitious reworking of the Power Macintosh 9500, but came at time when Apple was assumed to be circling the drain, and sold poorly.
Apple introduced the 1U rack-mountable Xserve in 2002, returning to a market segment it had bailed out of in 1996, following the complete failure of the Apple Network Server to establish itself. Xserve, as I wrote at the time in another publication, put Apple back on the list of serious server manufacturers. Although the fans of the first Xserve were deafening, and I wasn’t terribly excited about the integration of the server’s case with the rack-mounting hardware, I saw a great deal of potential in the combination of hardware and Mac OS X.
When Apple backed up its enterprise play in 2003 by adding the Xserve RAID to its lineup, I was again impressed. Although the RAID unit was best suited for shops that already had some experience with Fiber Channel networking, it was easy to set up and manage, and was supported by Apple with both Red Hat Enterprise Linux and Windows Server 2003.
Apple pitched the Xserve to academia as well as industry, with the stripped-down Xserve Cluster Node being aimed specifically at university environments. But outside of some very limited deployments in the film and video production industries and some HPC experiments, the world at large never fell for the hardware to the same degree that many of us in the press did.
Apple continued to refresh the Xserve hardware specs from one year to the next, by moving from PowerPC G4 processors to the G5, then transitioning to Intel Xeon CPUs from the Woodcrest, then Harpertown and ultimately, the Gainestown series. But when Apple discontinued the Xserve RAID in 2008, the first bell rang for me, and I spent much of this year wondering if there was one more Xserve in the works.
The company’s not completely abandoning the server market: it will continue to offer the Mac Mini Server, and will support Mac OS X Server on Mac Pros. But I have a hard time believing that anyone wants to waste 13U of rack space on two Mac Pros in a side-by-side setup, even in the 12-core configuration that Apple offers at the very top of the Mac Pro build options. The Mac Mini Server is attractive for small office deployments, being quiet and capable of supporting up to 50 users.
What remains a question is the future of Mac OS X Server; last month’s sneak peek at Mac OS X 10.7 “Lion” didn’t say a word about any plans for a server version of the OS. I have a bad feeling about this silence, but I hope I’m proved wrong.
Will Apple ever again try to play in the “Big Iron” arena? I kind of doubt it; in part because the market for enterprise-class hardware is far smaller than the consumer space, which the company seems to own for the foreseeable future. But I wonder if the knowledge that his crew has failed once again in the server market might encourage Apple CEO Steve Jobs to have another go at it someday.