IBM System z Solution Editions cut mainframe costs

The mainframe is way too expensive and IBM knows it, especially as a new generation of x86-based systems with robust virtualization emerges. IBM’s response is to roll out a series of hardware/software discount packages, referred to as System z Solution Editions. They promise to deliver the System z at apparently bargain basement prices. Here is a recent Solution Edition announcement from Aug.

In November, IBM briefed industry analysts on three more z Solution Edition offerings: for Chordiant CRM, Enterprise Linux, and Enterprise Linux Server. They join System z Solution Editions for Cloud Computing, Data Warehousing, SAP, ACI, WebSphere, Security, GDPS, and Application Development.

According to IBM, the z Solution Editions represent a “strategy to make System z attractive to new workloads so customers can run a much wider range of their business activities.” But the key to the strategy is slashing the price. System z Solution Editions are “designed to make the cost of deploying new workloads on System z price-competitive with alternative distributed systems,” says IBM.

For eligible buyers, which are those considering new workloads or are new to the z altogether, the Solution Edition bundles System z10 hardware, software, prepaid maintenance, and the necessary middleware at a substantially reduced price.

How much lower? IBM doesn’t reveal actual dollar-prices, preferring to talk in percentages. In one example IBM compared a customer running an SAP Solution Edition on a System z10 EC (6 GP engines, and 4 zIIP) to SAP running on HP (HP rx8640, 4×20 core 1.6 GHz Itanium servers). The result: 30% lower TCO over three years, 270% improved SAP throughput with the z SAP Solution Edition.

The Chordiant Solution Edition includes either a System z10 BC or EC, maintenance, and a software stack consisting of DB2 for z/OS and WAS for z/OS. It does require the company buy the Chordiant software too, naturally.

The Enterprise Linux Server Solution Edition, essentially, is a System z10 BC or EC machine with 2 IFLs packaged as an appliance for large Linux workloads. The entry configuration includes two zCores, 16 GB memory per core for the BC appliance or six zCores for the EC, 3-5 years of hardware maintenance, z/VM with 3-5 years of subscription and support, three 4-port FICON cards, and two 4-port OSA cards. IBM insists the acquisition cost is competitive with x86 or UNIX alternatives configured for large scale Linux consolidation.  Without seeing the actual costs or speaking directly with customers, however, it is not possible to determine how good a deal this is.

Along with the z Solution Edition pricing, IBM also cut the price of IFLs, as low as $50K per IFL in volume configurations. In the same spirit of price slashing Red Hat and Novell (SUSE Linux) reportedly are discounting their zLinux prices.

To add to the price incentives, IBM also launched a zRewards program last spring. zRewards  provides discounts on various migration services to help companies move workloads from other platforms to the z.  (An upcoming piece  will look at System z workload migration.)

So, is this going to be enough to persuade companies to move more workloads to the System z? IBM reports a 62% increase, 3Q07 to 3Q09, in the number of IFL engines shipped. The number of IFL MIPS shipped increased 100% over that time. This, however, doesn’t really prove anything about actually winning new customers.

The z Solution Editions certainly are a step in the right direction as far as lowering the cost of mainframe computing. The System z often makes a good TCO case for core enterprise applications, and its reliability and security are generally unquestioned. The problem, rather, lies in the cost of acquisition. To attract new customers, the System z needs a competitive cost of acquisition too.

Are System z jobs really available?

“Some large firms have reopened their college hiring” reports Alexei Miller, EVP at DataArt, a software outsourcing firm. “They’re actually talking about shortages of human capital.”

That apparently is the view from Canada too: “Demand for mainframe specialists is at an all-time high,” according to IBM Canada in a report published by The Industry Standard in Sept.

And openings for System z people are appearing almost daily on various listservs and forums aimed at mainframe professionals, particularly the mainframe and System z groups on LinkedIn

It is ironic that the big financial firms, for example, slashed IT staffing and, now that the markets appear to be rebounding somewhat, suddenly have projects they need to complete fast. And when they try to call those people back, Miller reports, many already have been picked up by others and aren’t available.

These projects, he notes, typically involve complex, non-standard work and massive data sets. The routine mainframe project work has long been shipped to India and elsewhere offsite. (DataArt also ships work offshore.) Rather, these are the kinds of projects that deal with pricing, risk, and compliance.

Yet unemployment has hit records highs and is predicted to go higher still.  In October, the unemployment rate rose to 10.2 percent, the highest since April 1983, according to the US Bureau of Labor Statistics. There seems to be an information mismatch here.

One underemployed System z professional searching for a full time position while frantically scrambling to complete System z contract work sent me the following lament:

Periodically, I am reminded that at this frantic time in my life there are four ways to catch up with one’s tasks at work.

I: Make a to-do list of tasks for the day, and during the day other tasks arise. At the end of the day, I find I have done most of the tasks for the day but some remain.

II: At the end of some days I have completed only some of the tasks and, naturally, some of my to-do list remains.

III: Other days as other tasks demand attention I find that I will do some tasks that day but none of the tasks on my to-do list.

IV: Then there are the days I end up NOT completing any of the tasks on my to-do list and NOT completing any of the tasks that came up during the day either.

An interesting aspect of this is that when I describe these stages to others, some find them hard to understand – especially how I could get to IV. Others will immediately recognize this from their own experience. The dividing line seems to be around just how much open-ended multi-tasking the job calls for. Knowledge workers who set their own goals understand; administrators who go from clearly-defined task to clearly-defined task don’t. Sales people understand; line workers don’t. First-level managers understand; plumbers don’t.

The problem is that many IT jobs are set up so that we all must fail at something. So, I try to pick what I will succeed at and hope.

NEON’s zPrime lowers the cost of mainframe computing

There is no small amount of FUD being thrown around the issue of NEON’s zPrime and the System z.  No doubt IBM and mainframe ISVs are unhappy. Mainframe customers, however, stand to save a lot of money.

At this point NEON, a System z ISV, finds itself countering IBM scare tactics, which target mainframe customers and suggest that the use of zPrime will somehow violate their IBM licenses. IBM, however, “has provided no specifics about any violations,” said NEON CEO Lacy Edwards in a recent phone call, adding “we’ve had lawyers from 200 customers review their IBM agreements and found no problem.”

zPrime is a technology that allows a mainframe customer to run almost any application on a zAAP or zIIP specialty processor. It can be used with all the core mainframe workhorse applications; IMS, DB2, CICS, even batch workloads. The product enables companies to shift these workloads to the two specialty processors, which reduces the workload on the general processors and typically result in substantial software licensing savings.

Since June, zPrime is under evaluation by 200 companies. NEON allows companies to put it into production before they buy it. Currently, 15 companies are evaluating the software in production.

zPrime is licensed based on MIPS like conventional mainframe software. For smaller shops (2000 MIPS) the cost would run about $300,000 annually. A large shop (100,000 MIPS) could pay as much as $5 million. Of course, as with everything having to do with IT, especially in this economy, prices are always negotiable.

For the System z customer the savings come from reduced charges for software running on specialty processors and from reductions in IBM’s licensing costs. Additional savings come if the increased use of the specialty processors eliminates the need to upgrade the system. In fact, depending on the applications and workloads, some customers may be able to drop down to a lower level of MIPS.

“We have one customer evaluating zPrime, a credit card processor with 30,000 MIPS, who estimates as much as $30 million each year in savings,” Lacy reported. Until independent IT analysts like me get a chance to speak directly to zPrime users, there is no way to validate the reported savings. Given industry mainframe software and hardware licensing practices and prices, however, reports of big savings seem feasible.

The winners and losers around zPrime are obvious: IBM and, to a lesser extent, mainframe ISVs no doubt stand to lose revenue. Customers, however, win big, bringing the cost of mainframe computing down to more manageable levels.

The mainframe is overpriced, and IBM knows it. That is why the company is doing everything it can to lower the cost of System z computing short of simply slashing prices outright. The latest attempt is the new Solutions Edition program. Similarly, IBM’s Open Infrastructure Offering (OIO) gives customers a way to aggregate various buys to get lower prices. Heck, the company has even offered IFLs (which don’t work with zPrime) at half price.

IBM has done very good things with the System z to revive mainframe computing and make it almost truly competitive. If products like zPrime can drive down the cost of mainframe computing further and can stop companies from abandoning the mainframe for distributed systems running the latest Intel multi-core processors or maybe even attract new customers and new workloads to the z, NEON will be doing  IBM and everybody else a favor.

BI returns to the System z

IBM’s Information on Demand (IOD) conference, recently held in Las Vegas, looked like a coming out party for business intelligence (BI) on the System z. At the event IBM unveiled a slew of products relating to business analytics and data management, many of which involved System z.

The party actually kicked off in June when IBM announced a smorgasbord of new or enhanced System z product offerings covering everything from application development to transactions to system management. Tucked into the announcement were a handful of BI-related announcements including InfoSphere Warehouse for z, Cognos 8 BI for Linux on z, a new version of DB2 for z/OS, and the coming of IMS 11, the latest rev of that venerable transaction database.

Some sessions at the IOD conference apparently turned into quite an IMS love fest according to attendees. Several large companies boasted running 200 million transactions per day on IMS 11, another combined IMS 11 and WebSphere in a SOA application. Yet another demonstrated IMS 11 running on the Amazon cloud—another case of decades-old mainframe technology showing off new tricks.

The conference actually highlighted what looks to be the second coming of BI on the z, a set of old mainframe capabilities long AWOL. The resurrection of BI on the z was a natural consequence following IBM’s acquisition of Cognos and the growth of Linux on the z.

BI on the mainframe once was the only way to get to the organization’s primary data, and it took the form mainly of static reports. With the rise of distributed systems, BI moved to those more dynamic platforms. Yet the data still resided primarily on the mainframe and often cumbersome steps were required to get it to the distributed platforms that would dice and splice it for BI purposes. This gave rise to data warehouses, data marts, and assorted middleware. The return of BI to the mainframe simply brings those capabilities close to where the data originates—in IMS, DB2, CICS and the other core mainframe systems.

What’s encouraging about IBM’s born-again interest in BI on the mainframe is the way the company is extending BI beyond the usual analytics. For example, it is taking unstructured content and making it available for BI analytics through tools like Cognos Content Analytics and InfoSphere Content Assessment. It also is incorporating unstructured content into master data management.

Finally, it is bringing mashups to BI. Mashups are browser-based end-user applications that combine data from different systems on a single screen.  IBM has offered its Mashup Center as a development platform for some time. Now it is adding the Cognos 8 Mashup Service to produce BI mashups.

Wells Fargo is one of the early mashup BI users. According to IBM, the bank is using mashups to pull together various pieces of existing information such account balances, the status of wire transfers, and real time market updates. You can bet a lot of mainframe data goes into those mashups.

So when it comes to BI, the System z is now back as a serious player. But this clearly is not your father’s BI.

By the way, I will be looking deeper into BI on the mainframe in the coming months. Here is a recent BI case study of mine, Univar. If your organization is doing anything BI-like that involves the mainframe and you’d like some glory, just let me know, alan@radding.net. I’d like to write about it.

System z hiring prospects

The job prospects for mainframe people today sound like something from Dickens’s Tale of Two Cities: It was the best of times, it was the worst of times…it was the spring of hope, it was the winter of despair…

OK, maybe “best of times” is pushing it given some of the discouraging talk recently on mainframe forums like LinkedIn. “My latest batch of interns is having a hard time getting places in mainframe. Could be the economy, because this has not been an issue to date,” posted a computer science professor recently to a job opportunity discussion. You can follow the discussion here. Still, Dickens’s words pretty much describe both the short- and long-term mood for those specifically interested in the mainframe and System z.

IBM’s latest corporate financials sound downright rosy, especially given the still dismal economy:  “We are optimistic about 2009 as we again raise our full-year expectations and we remain well ahead of pace for our 2010 roadmap…,” said Samuel J. Palmisano, IBM chairman, president, and chief executive officer in the company’s Q3 financials released Oct. 15.

And just to underscore Palmisano’s optimism a recruiter posted this seemingly attractive job opening on a different LinkedIn mainframe forum: Looking for a Senior Mainframe Developer to join a large and growing project with a long-standing client in our Baltimore office. The project provides much needed IT systems services in the Medicare community utilizing Mainframe technologies and has high visibility across the country for the 65 million users of Medicare. The full posting is here.

Beyond any immediate job postings, there have been much longer running discussions across multiple forums on LinkedIn and Facebook, around the future of a mainframe career in general. These are just more variations of the widely discredited  mainframe-is-dead question.

Another variation started cropping up several years ago. Data center managers would report they had numerous mainframe veterans poised to retire and didn’t know how to replace them. IBM began its Academic Initiative to address this by stimulating the cultivation of mainframe skills through university computer science programs. Today IBM boasts of over 600 participating schools. The recent grads I interviewed ended up with nice jobs at mainframe ISVs.

The economic downturn, which savaged 401k plans everywhere, discouraged a number of those mainframe veterans from actually retiring and averted the immediate crisis. Still, the problem seemed real enough. “At one point we couldn’t get enough mainframe application people,”  Kenneth Tye, CIO, TSYS, recently told me. Georgia-based TSYS is a leading global financial processing service provider.

TSYS solved its mainframe talent shortage by partnering with a local college, Columbus State University, even before there was an IBM Academic Initiative. Today, it doesn’t have trouble finding and retaining mainframe application people.

A problem, however, still exists when it comes to mainframe systems software people, the folks who work with z/OS, CICS, IMS and such. TSYS has been addressing this need, in part, by internally training mainframe operations people. Fortunately for TSYS it doesn’t need hundreds of these people, handfuls will do.

If you are an unemployed mainframe person, the picture must look glum. Wish I had consoling words to offer. Weak as they are, here’s the best I can do:

Mainframe jobs are available and more are coming. The TSYS mainframe environment is growing. The company started working with Linux on z and hopes to roll it into production in 2010. TSYS already employs 3400 people in IT, 85% of which are involved with the mainframe, and the company intends to hire more in the coming year.  Better yet, TSYS won’t be the only company hiring.

 

A competitive IBM System z market

Most of us have never experienced a competitive mainframe market. IBM controls the hardware and the operating system. It allows ISVs to provide applications under apparently pretty restrictive conditions. But that may change, or maybe it already is changing.

Once there actually was a somewhat competitive mainframe market when a group of vendors offered what was referred to as plug-compatibles, machines that would run IBM mainframe software without modification. These vendors were collectively referred to as the BUNCH, an acronym for Burroughs, UNIVAC, NCR, Control Data Corp., and Honeywell. With the departure of the BUNCH, not long after I started writing about mainframes, IBM completely controlled the mainframe market. Another competitor, Amdahl, now a subsidiary of Fujitsu, built rival large enterprise systems but never was a plug-compatible player.

In a lawsuit filed by T3 Technologies with the EU  the company complained of anti-competitive, monopolistic IBM practices. In announcing the filing in Jan, T3 cited an IDC report saying IBM controls over 99% of all existing IBM-compatible mainframes in use today. T3 apparently aims to build systems for small data centers, 500 MIPS or less.

Spurred by complaints from T3, the US Justice Department reportedly has launched a review of IBM’s competitive practices. This may account for recent IBM activities that otherwise seem baffling.

For example, IBM seems willing to tolerate Hercules, an open source software implementation of the mainframe System/370 and ESA/390 architectures and the new 64-bit z/Architecture. Hercules allows you to run mainframe applications under Linux, Windows (98, NT, 2000, and XP), Solaris, FreeBSD, and Mac OS X (10.3 and later) as an emulation.

OK, I can understand IBM’s reluctance to crush Hercules. After all it is emulation and open source so it probably won’t make much of a dent in IBM’s revenue. Some of the OpenSolaris on z people even look to it to test their efforts on an x86-based system since many lack ready access to an actual System z.

But now there is TurboHercules, a commercial version. Suddenly there is money at stake. IBM probably has not cracked down on TurboHercules because the company exploits a tiny loophole in the IBM mainframe license that allows a licensee to run its software on a different platform in the event its mainframe is temporarily unavailable. For organizations with only one System z, TurboHercules would enable them to legally recover their systems temporarily on another platform, say a large Intel box.

So what explains IBM’s tolerance for Neon’s zPrime, which helps more mainframe applications to run on System z specialty processors as a deliberate strategy to lower software licensing costs? ISVs tell me that they could do what zPrime does but they have refrained out of fear that IBM will crack down. They are just waiting for IBM to pounce on Neon.

In truth, IBM has been schizophrenic about openness and control for years. On the one hand IBM zealously guards its System z intellectual property. On the other, it created the open source Eclipse community and contributes code. More recently, it has called for an open, standards-based cloud environment.

Maybe the sniffing by the Justice Department is restraining IBM’s most restrictive tendencies. And in this little window, TurboHercules and zPrime just might take root.

The bigger question is whether the mainframe actually is a monopoly today. Customers, usually smaller companies, migrate off the mainframe every year for distributed platforms from HP and Sun. So even if IBM controls the mainframe, it doesn’t control enterprise computing.

There are customer advantages to IBM’s control of the System z environment in terms of investment and innovation, not unlike what Apple customers gain from its control of the Mac platform. Yes, System z prices remain high, but IBM, faced with x86-based competition, is scrambling to reduce customer TCO every possible way without actually cutting z acquisition prices. Competition is good.

Has the System z become hot?

IT pundits continue to argue that the System z is a dinosaur heading to extinction (hence the title of this blog, dancingdinosaur).  The Register, a UK publication, implied that in a recent article about the System z unable to compete on price/performance against the latest Intel Nehalem processors discussed here in September.  Several mainstream business and IT publications said the same thing in articles citing the Union Pacific Railroad’s slow, painstaking, and not yet completed migration from the System z to servers running commodity processors.

The truth: the System z ain’t heading for extinction. Even Union Pacific may never get its critical workloads off the z. And if it does, that migration may signal its last stumbling steps toward its own extinction.

A number of recent studies from CA (reported here previously), BMC, and IDC, a source with no more stake in the z than in any other platform, all point to the System z weathering the current economic turmoil and to signs for a brighter future going forward.

The latest survey by BMC, the fourth in the past four years, shows the System z chugging right along with 56% of respondents expecting to increase their MIPS, and this came despite a declared priority to reduce overall MIPS growth, typically as part of a consolidation and cost savings strategies. Seventy-five percent of larger mainframe shops expect to increase MIPS.

A very small number of shops (11%) reported plans to consider leaving the System z altogether, but these generally were small shops whose workloads may indeed be better suited to other platforms. A slightly large number of shops (12%) reported plans to consolidate more workloads on the mainframe. Most (77%) expect to keep their workloads right where they are.

More significant findings have come recently from IDC (International Data Corp., Framingham, MA), which is a platform neutral IT research firm, unlike CA or BMC, which have vested interests in mainframe growth. As reported by Reuters in mid-September nearly one-half of 300 survey respondents reported plans to increase annual spending on mainframe hardware and software.

The IDC study describes the emergence of a blended workload approach for the z. According to Jean S. Bozman, research vice president with IDC’s Enterprise Platforms Group:  “Customers are finding that new workloads, including Linux-based and Java-based workloads, can leverage the mainframe’s built-in security and high levels of availability, by running them on mainframe specialty processors, such as the IFL, zIIP and zAAP processors. This pattern of adoption is placing software licensing costs on a lower price schedule for these new workloads than if they were running natively on the IBM System z hardware platform.”

This ability to shift workloads to specialty processors is the key to significant savings in software licensing costs. And the number of workloads organizations can shift to the z is poised to increase dramatically if Neon’s zPrime gains traction. zPrime enables companies to run IMS, DB2, CICS and batch legacy applications on specialty processors, thereby reducing mainframe computing costs by as much as 20%, according to Neon. (Dancingdinosaur will have more on zPrime in upcoming postings.)

As note previously here, with OpenSolaris on z and .NET applications via Mono there is the possibility of greatly increasing the workloads able to run on the z, making it an increasingly cost-efficient platform. Add in the cost savings possible through running a growing assortment of applications on specialty processors and the z doesn’t look at all like a platform heading for extinction.

More Windows on the System z

There are no announced efforts by either IBM or Microsoft to enable Windows applications to work on the System z. Got it? None, nada, nothing…

Yet, the subject keeps coming up. The most recent to take it up come from a couple of Windows publications,  of all place. Both Redmondmag and Redmond Developer published the same piece on the topic, Windows Coming to a Mainframe Near You? by Stephen Snoyer, a New York-based freelance writer.

Similarly, no one except possibly me is offering a plausible business case for Windows on z. It seems that the various Windows on z efforts like Mantissa’s v/OS and WINE are occurring more for the technical or intellectual challenge than for any serious business need.

IBM continually scoffs at the idea of Windows on z. They will give you technical reasons, raise legal issues, and cite business concerns. None appear insurmountable, and even the various IBM spokespeople don’t seem convinced by their arguments.

The best stab at Windows on z has been by Novell, which offers Mono through SUSE Linux running on System z. Mono, as noted here back in early August, is a complete, up-to-date Linux (open source) implementation of an ASP.NET application server that runs on the x86 and System z platforms.

In effect, Mono is a virtual machine for .NET. Using Mono running on SUSE Linux on the System z, organizations can bring their .NET workloads onto the System z.  This is not exactly Windows running on the z. While .NET applications usually are Windows applications, not all Windows applications are .NET. So, not every Windows application is a candidate for Mono; only those that are .NET apps and conform closely to the .NET standard. And even then there may be a certain amount of tweaking required.

The real reason so little has happened to move Windows onto the System z is the lack to date of a compelling business case. In addition, data center managers don’t see it as a battle worth fighting. The stars, however, may be aligning for a business case. In short, it is the same as the case for consolidating Linux workloads on the System z.

Enterprises with System z data centers also have tons of Windows and .NET applications pervasive throughout their organizations. Many of these are not even visible to the IT organization. Rather, they are small workgroup and departmental applications running on a stray server with attached storage, maybe even an old PC that has been demoted to serving a single Windows application for a handful of users.

If, and this is a very big if, the organization can smoke out these applications and consolidate them on the z, there appears to be the potential for significant improvements in management efficiency through consolidation and increased application reliability and security through the System z. This, essentially, is the same pitch for Linux on System z.

In this case, the technical issues probably pale in comparison to the political obstacles involved. Users won’t give up control of their Windows applications easily. However, if the savings are real, there should be budget to bribe them to go along.

Oracle vs. System z with Linux

The battle between Oracle and IBM over the future of the Sun SPARC/Solaris customer base appears to be getting testy judging by a recent ad that ran in the Wall Street Journal on Wed., Sept. 9. The ad reads: “We’re in it to win. IBM, we’re looking forward to competing with you in the hardware business.” signed, of course, by Larry Ellison, Oracle’s top dude.

Huh? Oracle and Ellison have never competed before in the hardware business against anyone. And Sun, which Oracle recently acquired, wasn’t doing very well competing against anyone—HP or IBM—before the acquisition. So I’m not sure Oracle, a newbie to the hardware business, is going to do all that well, at least not in the short term.

Meanwhile, IBM has been racking up wins against Sun for several years. In the first half of 2009, IBM doubled the amount of Solaris migrations to Linux on IBM servers.  In that period, IBM reports over 250 Sun wins. Of course, not all the Linux migrations to IBM have gone to the System z, but some have; 40% of new z customers installed Linux in 2008, according to the company.

Just last May IBM launched a new effort to grab Sun accounts with a package of price cuts and migration services to make it cheaper and easier than before to move from Sun to System z. You can see the announcement here.

In addition, IBM has been raiding Sun ISVs. The company claims 70% of ISVs are interested in a closer relationship with IBM and half of those are already actively engaged in deeper technical enablement or go-to-market activities. Kabira Technologies appears to be IBM’s latest trophy ISV. In addition, IBM reports 91% of Sun resellers are interested in a closer relationship.

The Oracle counter punch seems limp. The company promises to spend more money developing SPARC and spend more developing Solaris than Sun does now. But Sun has been so hobbled financially, it is not clear if Sun was spending much at all recently.  Similarly, Oracle is promising twice as many hardware specialists selling and servicing SPARC/Solaris systems than Sun does now. Again, this may not be such a big number. It wouldn’t surprise me if IBM has almost as many people chasing Sun accounts as Oracle does.

Oracle, however, does promise dramatic improvements in Sun’s hardware performance by tightly integrating Oracle software with Sun hardware. Give this one to Oracle. The only question is how long until the necessary tweaking is ready. Don’t bet on performance improvements showing up real soon.

Finally, there is the OpenSolaris on System z issue. Efforts on this continue although they have been sputtering of late due, in part, to the upheaval at Sun. Nobody expects Oracle to give this effort much attention. Heck, even IBM barely gives it the attention it deserves. When that happens, however, it will be one more reason to move from SPARC/Solaris to z, even if Solaris and OpenSolaris are not quite the same.

System z price cutting

IBM has quietly cut mainframe prices for years. With the economy in the dumps and x86-based chip vendors bringing out new high end processors, however, IBM can’t be coy about price cutting or risk losing deals to lower cost x86-based platforms.

The price cutting suddenly became quite public when The Register, a UK publication, ran a piece on IBM slashing IFL prices in half in the response to Intel’s latest chip, Nehalem. The article provided a detailed cost analysis of the System z versus comparable workloads running on systems running the latest Intel chip.

The analysis concluded that for the same money,  “a fully-engined System z BC with five 3.2 GHz engines would need to support the same number of Linux images as 48 Intel-based systems running a total of 384 Nehalem cores running at 2.93 GHz.” At a minimum, that would translate into 384 Linux images.

The writer implies that the System z price/performance can’t match the newest generation of x86-based servers when running Linux workloads. I have not seen an IBM response to the article, but the writer’s cost analysis falls into a common trap. He compares initial acquisition costs and throws in some five-year support charges. That’s not nearly the whole cost story.

DancingDinosaur has covered this issue before, in March. The problem is that there are more costs to consider than just the hardware, software, and maintenance.

Three key cost issues were overlooked in The Register’s analysis:

1. The System z is a more robust, secure environment. It was designed from the start to be industrial strength. You can aggregate several hundred Nehalem cores, but you will have to build the industrial strength attributes into the resulting system on your own if you hope to bullet-proof the system. That won’t be easy or fast and certainly won’t be cheap. You’ll have to add the cost of that to your calculus or add money to offset the increased risk, which will be significant.
2. The people issue. Mainframe data center managers run lean shops. They don’t need a lot of people to handle the care and feeding of the System z. Shops running systems comprised of large numbers of commodity processors generally need considerably more people to keep those systems running at optimum levels. And while the cost of hardware and software goes down, the cost of people only goes up. If you want to run a cost-efficient data center, you have to minimize your people costs.
3. The cost of energy. Admittedly, I don’t know what the energy consumption of 384 Nehalem processors comes to. The only figure I found puts the latest Nehalem delivering 10x the power while consuming 3x more energy than the previous generation. There were no comparisons to the System z, but somehow I doubt it is less than the System z running a comparable workload.

A recent survey by CA of executives at System z shops found 76% reporting that managing a large number of distributed servers in their data centers has become a cost issue. In addition,  67% noted that as the distributed infrastructure grows, the ability to run multiple applications on a single mainframe becomes more attractive. Hmmm.

So, don’t accept The Register’s cost analysis as the last word. Let’s see what the next System z processor rev brings.