In-Depth

Q&A: Big Iron’s Big Resurgence No Accident

A senior IBMer notes how the company has made the mainframe a more affordable proposition for traditional and non-traditional customers alike

With five quarters of successive growth behind it, and with new ISVs clambering on board every year, the mainframe is stronger than ever. Colette Martin, zSeries program director with IBM, spoke with Enterprise Strategies about mainframe market trends. Big Iron’s comeback is no accident, says Martin, who stresses that IBM has done a lot to make the mainframe a more affordable proposition for traditional and non-traditional customers alike—and that Big Blue may yet have more tricks up its sleeve, too.

Earlier this month, Informatica—a relatively high-profile vendor—introduced its first native application for z/OS, an interesting move because Informatica targets a relatively compute-intensive workload—ETL processing. Rightly or wrongly, there’s a perception that workloads of this kind are more expensive to do on the mainframe than elsewhere. So was this something of an anomaly, or are customers starting to deploy more compute-intensive workloads on the mainframe?

The things that customers have traditionally been running on the mainframe, they’ve continued to run on the mainframe. And this has actually grown over time. The dynamic we’ve seen there is that as customers do more new workload processing on the mainframe or elsewhere, it tends to drive more need for some of the data transaction-oriented processing that drives really where the mainframe’s strengths traditionally have been.

It’s good for the mainframe, it’s good for the IT industry. We’ve really got this core of technology that customers are really positioned to be able to leverage now, where they move more toward becoming on-demand businesses.

One of last year’s highlights was the availability of the long-promised z/OS Application Assist processor. I believe you announced it in the summer [of 2004], but didn’t ship it until later that year. Since it’s been available, what kind of interest have you seen from customers?

zAAP really became available in Q4 with z/OS 1.6, and we have seen an enormous amount of interest in the zAAP and customers moving to zAAP implementations. It’s still a little bit early, but in terms of a new technology intro, it has succeeded any of our internal expectations in terms of acceptance in the marketplace. The good news there is that we’ve given customers a way to very easily and affordably implement Java-based workloads like WebSphere on the mainframe, which before zAAP wasn’t always the case.

So you’re now seeing more customers running WebSphere natively on the mainframe, too?

WebSphere is absolutely one of the most popular new workloads [for the mainframe]. Whereas in the past, the choice they had on the mainframe was to run [WebSphere and other J2EE workloads] on z/OS on a standard engine, they were then being charged for all of those MIPS on a standard engine at standard engine prices.

Now they have the ability to run half the workload on the zAAP engine and the other half on the standard engine, and there are no IBM software charges on that zAAP engine. The cost of the engine itself is relatively inexpensive—both the [Integrated Facility for Linux] and the zAAP engines are very affordably priced from a hardware perspective. And the software—there’s no IBM software charges on the zAAP engine, and so MIPS they might have needed to use on the standard engine before can now run on the zAAP engine, and the price performance on that is dramatic.

You’ve done a lot to make the mainframe a friendlier place for Linux-based workloads, what with your aggressive IFL pricing, and now for J2EE workloads and WebSphere, too. But what about the often prohibitively high cost of native z/OS workloads? Any progress there, in terms of making the mainframe and z/OS more affordable for traditional and non-traditional customers alike?

I think the new engine pricing, as well as things like sub-capacity pricing, have all done very positive things for our customers. We’re seeing more and more ISVs that are supporting sub-capacity pricing, in particular, and this can make the mainframe a much more affordable option for many [customers].

The other thing that we’ve done over time is customers are finding that if they stay current on the latest technologies, they do see significant price/performance improvements. They see their software pricing go down, and they see the capacity they’re getting for the same price going up. There have been a number of initiatives really aimed at helping customers grow those new workloads more effectively and affordably on the platform.

One interesting thing here was your z/890, the mid-market mainframe, which, I believe, is targeted at a relatively new market segment for the mainframe. What kind of reaction have you had from customers to the z/890 since its introduction? And what did you change [in the z890] to make it a more affordable proposition for mid-market customers?

That z890 program has been by far the most successful 390 or zSeries program we’ve ever introduced. And one of the things we did last year when we introduced the z890, we introduced a lot more choice and flexibility in terms of the size and number of engines a customer could have on the z890. That’s also helped customers move forward more affordably on the mainframe, because they can pick the size machine that’s just right for them. Not only that—once they’ve moved into the latest zSeries technology, their opportunity for growing is much more flexible and easy, because they can grow in smaller increments. So instead of having to bite off a 300 MIP chunk [of capacity], they can move to something that’s much, much smaller than that if the thing they want to implement is a tiny workload.

I understand it’s possible to purchase z890 machines with less capacity than even the “Baby Z” z800 mainframe. What was your strategy there? Why not do the same thing with the Baby Z?

Well, yes, it is possible. On the z800, when we first introduced it, the smallest entry point was 80 MIPS, and we actually introduced, maybe six months later, we introduced a 40 MIPS model. The z890 we actually started at 26 MIPs, so it’s just a little larger than half the size of the smallest z800 that we had. While those very, very smallest customers are typically not the ones that are growing, it does give them a way to continue to maintain the mainframe workload they had.

In some cases, it’s something they had running for many, many years. It’s something they want to keep running on the mainframe because they know it’s the most available, most secure, and they want to continue to have that level of support. So we’ve given them a choice to be able to move that forward.

So notwithstanding its mid-market focus, the z890 is also aimed at older customers who just want to maintain the [mainframe] capacity they have?

A customer who’s perhaps sitting on older hardware and needing to make a choice on a given point in time where they go, they can stay on the mainframe and most of them want to stay on the mainframe. With [the z890], they can grow very affordably within the family by just adding a second about-26 MIP engine. Of course, as you add additional engines, there’s a little bit of a performance effect, so they’re not exactly 26 MIPS each as you add the additional engine. But the point is this gives them a way to incrementally grow their capacity.

One hallmark of the mainframe resurgence has been the number of new ISVs introducing mainframe applications for the first time, or re-entering the mainframe software business after having abandoned it years ago. We saw this trend in 2003, and I heard encouraging reports about 2004. What happened last year on this front, and do you expect this trend to continue in 2005?

We definitely have had far more interest [from new ISVs], but I don’t have more names to share off the top of my head. I know that there were several dozen new ISVs, and several times that in new applications, mostly for Linux.

There’s also a whole suite of [IBM] Software Group applications that are available to customers, and, in fact, for almost every ISV program that is out there, our software group colleagues have something similar—and they have very extensive [application] migration offerings as well, so customers have a choice. That gives them the flexibility to be able to negotiate [with ISVs], and that helps the ISVs be more positive in this scenario as well.

Any final thoughts you’d like to share on the mainframe’s resurgence?

I guess just this: As of the end of 2004, we had seen five quarters of growth on the mainframe. We’d grown share consistently throughout that time period, and we’re seeing dramatic growth in MIPS on the platform. We saw 32 percent growth in MIPS on the platform in 2004. That’s just phenomenal, and it is, in fact, a statement of the new workload acceptance on the mainframe as well.

And while there may have been lots of naysayers in the marketplace, the folks that were actually mainframe customers never really left. They’ve been our strongest supporters all along, and they’ve been right behind us as we’ve revitalized our strategy and are moving forward in the future.

Must Read Articles