Operating Systems

IBM was kind enough to solicit input from a number of analysts (including me) on the presentation and positioning of the z10 just prior to the February 26 release date. As usual for any System z release, there was a flurry of discussion of new hardware features and new z/OS-specific function. One of the things that really stood out in the discussion was the central role that virtualization—and specifically Linux for System z and z/VM—played in the actual announcement text and slides. The sheer shock of this great departure from traditional IBM “one operating system to rule them all” per platform policy was interesting in how much that tells us about where IBM has seen the most value from the Linux investment—in the virtualized world that 35 years of work has kept alive, making IBM moderately large chunks of spendable cash to fund useful stuff.

Given IBM’s occasional hostility toward VM and virtualization in general (Melinda Varian’s superb “VM and the VM Community: Past, Present, and Future” paper [www.princeton.edu/~melinda] is a terrific background read to IBM’s perennial love-hate relationship with VM), this change of heart tells me a few new things about the way IBM wants to focus its resources in the next year or two.

First, acknowledging a little help from the VMware on Intel crowd, the concept of a completely virtualized infrastructure is pretty much legitimized in the same way that the IBM PC legitimized the personal computer. IBM, in prominently featuring z/VM and Linux in the announcement of its largest and most powerful System z yet (and make no mistake, the z10 is a quantum leap in performance and functionality over even last year’s z9 BC and EC), has stamped the idea with the “ready for prime shift, no excuses” label. It will be interesting to see how marketing materials—and the marketing message—reflect this going forward. The traditional incentive for System z sales has been to maximize z/OS revenue; by putting VM and Linux workloads on a par with z/OS workload, and having the performance to back it up vs. heavily distributed environments, will change a lot of things about the way IBM markets its toys to us.

Second, we’ve received the green light to legitimize Linux as an “approved” companion to z/OS workloads. Several IBMers I regularly talk with tell me that the majority of z/OS customers now have the capability to support Linux applications, and the concurrent retirement of the 374x communications controllers and migration to Communications Controller for Linux (CCL) have put significant numbers of Linux systems into play in real-world data centers—enough to require IBM to start treating Linux compatibility as a mandatory checkbox for the announcement letter. It’s entertaining to think that when I was on tour in Canada talking about Linux on z in 2000 to 2001, people would look at me funny when I would answer the question of “Who’s ever going to run Linux with z/OS?” with “Every single z/OS system out there.” Less than a decade later, the head of the STG System z division of IBM is featuring Linux on the same stage as z/OS.

I could go on, but this sums up a very important step for IBM and for the Linux community supporting System z. It’s about time, and it’s about mindshare both in customer and in IBM brains. Putting this kind of effort and visibility upfront and center in the z10 announcement is heady stuff. I look forward to seeing what both communities can do with it. Z