Mar 1 ’06

z/Bottom Line: A POWER Trip

by Editor in z/Journal

Our beloved pooch Elle has always loved a good soap opera, and in our industry, there’s never been a better one than “The Days of Our Mainframes.” In the latest episode, the mainframe is once again called on to justify its reason for living. Z makes an impassioned plea that it continues to serve up the majority of the world’s data, and strives for continued acceptance amongst its, uh hmm, “peers.” Lately, there’s been a great deal of gossip behind Z’s back about not being cost-competitive (old rumors die hard). The mainframe, long on performance but short on friends, continues to try and blend in, and make itself just another pretty face in the world of computing. 

Just when you felt sorry for our much-maligned platform, there’s an important engineering and R&D effort under way by IBM design that will once again enhance and extend the value of the mainframe. 

It’s All About POWER  

The very processor that until recently powered Apple’s Mac computers, and whose derivatives will even power Sony’s upcoming PlayStation 3, is the focal point of a project within IBM that has been coded under many names, including POWER Everywhere, and the most meaningful for our world, “eCLipz.” As chronicled in this column and many other places in the last several months, the 40th birthday of the mainframe reminded us of the reasons for its origin. At the time, IBM was simply manufacturing, selling, and supporting too wide a range of computers. The S/360 was Big Blue’s answer to consolidate all technology into one family of computers. Fast forward to today, and IBM finds itself once again as the only company with all the platform answers; today known as i, p, and zSeries computers. But once again, the divergent efforts are limiting their approach to be as competitive as needed in today’s über-competitive world. 

It turns out that the eCLipz project, which according to consultancy Isham Research, most likely stands for “enhanced Core Logic for iSeries, pSeries and zSeries,” is aimed at delivering common hardware throughout IBM’s offerings. As Isham also points out, while descriptive, the eCLipz code name could also have much to do with covering up IBM’s competitor, Sun. 

The current POWER5 generation has already made tremendous strides in this area. You may be surprised to learn that OS/400 runs on essentially an unmodified pSeries system based on the POWER5 technology. This has yielded tremendous financial rewards for IBM by combining R&D costs shared across both very successful product lines. While the offspring of eCLipz, the upcoming POWER6 generation, will continue to deliver a common processor for i and p, the z/Architecture is just too different to be identical. But according to IBM roadmaps, portions of the POWER6 will be common across all three heretofore proprietary and separate systems.

 What this means for the mainframe industry is profound. First, it confirms IBM’s solid planning for evolving the mainframe, period. Second, the evolution calls for morphing the mainframe into more of a commodity product, distinct from its unitary standing today. This will mean lower manufacturing and R&D costs for IBM, which ostensibly should improve the affordability for customers in comparison to other platforms. And with the z/Architecture moving toward essentially common hardware environs, we could envision a data center with a farm of Unix/Linux, OS/400 and z/OS, z/VM and z/VSE systems (even all in one cage of Blade servers), connected to common storage capabilities (e.g., a Shark storage subsystem) all blended together as one happy family. The path of convergence will cause the computing landscape to blur—as it should be.

 There may also be some elements of hope for the low-end mainframe marketplace in the eCLipz deliverables. The vast success of IBM’s mainframe on a card, aka the P/390, showed how small shops and large shops with segregated development would take advantage of a relatively inexpensive mainframe. With the POWER6 platform leveraging many other commoditized components, we may see the grandchild of the P/390 in terms of entry-level affordability. 

And what about Linux for zSeries? Since Linux is fully supported today in the POWER environment, what will the advantages be to “host” Linux in a zSeries environment? Perhaps some features such as HIPERSocket memory-speed I/O between traditional mainframe and Linux workloads will still offer advantages, but that remains to be seen.IBM’s eCLipz strategy is on-target to leverage the best of all combinations to deliver more for less on the hardware side of the equation. Now IBM needs to find a way to align the software costs even more dramatically along usage lines, vs. capacity. And while all sources indicate we won’t see the results of the eCLipz project until late 2007 or early 2008, the POWER of the effort once again extends the importance and feasibility of this venerable platform. All this makes this one of Elle’s favorite episodes.

 And that’s z/Bottom Line.