In-Depth

Analysis: Why Big Blue's Power Systems Consolidation Matters

One expert explains why last month's Power systems consolidation is so important.

• By Stephen Swoyer
• 04/22/2008

At last month's COMMON user conference in Nashville, Big Blue unveiled two standard configuration systems -- the IBM Power 520 Express and the IBM Power 550 Express -- that can run the AIX, Linux, or IBM i operating environments (see http://www.esj.com/enterprise/article.aspx?EditorialsID=3100).

That was the big story, of course, but it wasn't the only story.

In fact, industry watchers say, IBM's System i and System p consolidation is proof positive of "sweeping" organizational change inside IBM's Server and Technology Group (STG). What's more, it constitutes a major expansion of Big Blue's virtualization portfolio, as well as a much-needed rationalization of its product naming and branding efforts. In other words, experts say, last month's Power systems consolidation is a Very Big Deal.

"[W]e've seen the most significant changes to the way IBM STG runs its business and [how it] presents itself to customers since 2000, when it corralled its historically fragmented server lines under a single organization and R&D structure, publically identified as the umbrella 'eServer' brand," writes Gordon Haff, a senior IT advisor with consultancy Illuminata.

The upshot, Haff concludes, is a more "streamlined" Power systems lineup.

"All servers run standard POWER6 processor modules, whereas the POWER5 generation included Quad-Core Module [QCM] variants offering more cores running at a lower frequency than the standard Dual-Chip Module, in addition to using a Multi-chip Module [MCM] at the high-end. The Power 595 now anchors the high end of the line by itself, where previously there was both a p5 590 and a p5 595," he points out. "Product lines tend to naturally get more cluttered over time in response to competitive announcements and customer requests, but IBM has made an honest effort to de-clutter the Power lineup."

There's also new -- and much-needed -- coherence on the virtualization front. Big Blue has used several different branding strategies to highlight (and frame) its Power virtualization technology, Haff points out; PowerVM is the latest and the most coherent, he argues. "PowerVM is IBM's latest effort to pull together and talk about its virtualization offerings in a systematic way. 'Virtualization Engine' was one earlier, less-than-wholly-successful effort," Haff concedes.

It's a timely move, he suggests. "[I]t's become more important for IBM to better organize its virtualization portfolio. Even as virtualization becomes more widespread and important to more businesses, it's also become more complicated," he continues, noting that Big Blue recently added the ability to move running LPARs from one server to another without first shutting it down.

"Live Partition Mobility is IBM's answer to VMware's VMotion," Haff says, and "IBM has also supplemented LPARs with AIX Workload Partitions [WPARs, or operating system containers] that can also be dynamically moved, though in more a shutdown, then quickly restart [kind of] way."

Indeed, Live Partition Mobility could be PowerVM's -- and POWER6's -- killer app. It delivers a capability that software-only solutions (such as VMWare) simply can't touch.

"IBM says the observed 'pause' for the shift [of a running LPAR from one server to another] is about two seconds -- within the TCP/IP timeout window, so that network clients won't even notice that the LPAR is suddenly running on another server," Haff points out. "[S]oftware-based approaches such as VMware make similar claims, [but the] reality [of these claims] is highly load-dependent.”

No top-to-bottom system reorganization would be complete without an attendant software rationalization. In Big Blue's case, a rationalization of this kind was long overdue. "IBM is taking a stab at rationalizing its Power Systems software as a whole, thus the new Power Systems Software group. In addition to virtualization and operating system/integration layers, IBM groups the software into availability, security, energy … and management [i.e., IBM Systems Director] buckets," Haff explains.

"Some early changes are mostly about creating a more consistent nomenclature," he continues, noting that Big Blue's "long-standing Unix clustering product, High Availability Cluster Multi-Processing [i.e., HACMP] … becomes PowerHA for AIX and PowerHA for Linux while System i High Availability Clusters [i.e., HASM] become PowerHA for i."

Haff and Illuminata anticipate that IBM will continue to reconcile development roadmaps and enhance p/i technology sharing over time, too.