Mainframe Computer

Putting it all together: PSI, the mainframe, and blade servers

When IBM announced that it was buying Platform Solutions Inc. last year, most people thought Big Blue was doing it to shut PSI up.

Before the acquisition, IBM and PSI were in a nasty legal fight over 1) whether PSI was violating IBM copyright by using z/OS on its Itanium-based servers; and 2) whether IBM was trying to monopolize the mainframe market by binding z/OS only to its own hardware. Now that fight is over, although T3 Technologies, a company that had a licensing and distribution deal with PSI, is continuing the legal battle, as documented by The New York Times in a piece this week (and as we wrote about in January).

Needless to say, there are some who think PSI had more to offer IBM than just lawsuits, and saying otherwise is unfair to former engineers at PSI and Amdahl (Amdahl being the company that spun off to become PSI). See, PSI’s biggest selling point was that its servers could run z/OS, Unix, Linux and Windows all on a single Intel Itanium-based box. So perhaps IBM is taking that capability and using it for its own mainframes and other servers.

In a follow-up blog to the Times story, Ashlee Vance wrote this:

The Platform engineers had also built technology that would offload some mainframe jobs like encryption and data analysis onto separate machines such as x86-based servers or even machines using I.B.M.’s Cell chip. The idea was to speed up these jobs with more modern hardware and to create a path between industry-standard servers and mainframes.

Such technology could well appear in new mainframe systems due to arrive in late 2010 or early 2011, according to numerous people interviewed for my story.

This could tie into the System z keynote speech at Share, which was given by Karl Freund, a VP in the IBM System z group. In an interview after the keynote, Freund divulged some details. Think blade server as a mainframe LPAR:

“It will be like being able to treat those blade servers as if they were System z, from a systems management perspective,” he said. “Extending the z role to a heterogeneous environment.”

According to Freund, it would be like running a blade server as if it were another logical partition (LPAR) on your mainframe. Though there is little comparison between the hardware of a mainframe and an Intel blade, Freund said management will be easier, being able to handle one systems management console, and having failover going to the same sysplex, rather than using different backup platforms.

T3 Technologies pushes forward with complaint of IBM mainframe “monopoly”

According to ZDNet, T3 Technologies filed a complaint with the European Commission on Tuesday alleging that IBM was abusing its power in the European mainframe market.

T3 is an alternative mainframe company and one of the only ones out there. Last year, IBM and Platform Solutions Inc., another plug-compatible mainframe vendor, settled their legal battles when IBM announced it was buying PSI, essentially shutting it up.

Since late 2006, PSI had been in back-and-forth lawsuits with IBM in which IBM sued PSI for patent infringement on its z/OS mainframe operating system, and PSI firing back with a countersuit claiming that IBM was shutting out competition by coupling z/OS with its mainframe hardware. The fight was settled with the acquisition, and since then, PSI has vanished. PSI’s web address, www.platform-solutions.com, redirects automatically to IBM’s System z mainframe page. And in IBM’s SEC filing earlier this year that explained the acquisition, Big Blue wrote the following: “PSI’s technologies and skills, along with its intellectual capital, will be integrated into the company’s mainframe product engineering cycles and future product plans.” Which is kind of like saying, “Yes. PSI is pretty much disappearing.”

A Gartner report on the acquisition said that Itanium-based mainframes, which is what PSI was selling, would go the way of the dogs, but that some of its engineering know-how, such as virtualizing I/O across heterogenous environments, could get worked into the System z brand.

Needless to say, that acquisition did not rid IBM of T3. When PSI sued, T3 filed its own motion in support of PSI. T3 had a licensing agreement to resell PSI technology under its Liberty Server brand, and planned to continue to support its existing customer base. Upon the announcement of the IBM acquisition of PSI, T3 Technologies President Steve Friedman said T3 was “not affected by the acquisition of PSI by IBM.”

That may have changed, though, as indicated in the new legal complain that was detailed in the ZDNet story:

IBM is accused of engaging in a range of anti-competitive actions, including “preventing the sales of competing mainframe hardware products by tying the sale of its operating system to its mainframe hardware”. IBM is further accused of “withholding patent licenses and certain intellectual property to the detriment of mainframe customers”.

In its statement, T3 said it sold IBM mainframes but then moved into selling its own products, the Liberty line of mainframes, first developed by Amdahl Corporation, which itself is now part of Fujitsu.

Will T3 go the way of PSI? When I asked Friedman this back in July, he said only that “at the moment, we’re continuing on as we have been and supporting our existing customer base.”

CA looks to “exploit” the zIIP

Mainframe software giant CA today announced a new version of IDMS, its mainframe database management system, that in particular takes advantage of the zIIP specialty processor on big iron to save users money.

The zIIP, which stands for z Integrated Information Processor, is a mainframe processor designed specifically to run database-like workloads. DB2 has been a popular choice for the zIIP. It joins other specialty processors like the Integrated Facility for Linux (IFL) and z Application Assist Processor (zAAP), which are geared toward Linux and Java apps.

One of the benefits of these specialty engines is that you don’t have to pay software licensing costs for the workloads that run there. That is probably the most direct savings you’re going to get.

But as I’ve written before, some of the real savings can be getting workloads off the central processors, because that can free up enough space there so that users can defer having to buy another mainframe. It’s that seven- or eight-figure capital cost of a new mainframe that can be the really big, possible savings when moving workloads to these side engines.

CA also announced that its Datacom software, another database management system, is in beta release and will also be able to run on the zIIP.

A mainframe on layaway, and other IBM incentives

In a move taken from auto dealerships and furniture companies, IBM is offering a way for mainframe users — or wannabe mainframe users — to get big iron now without having to pay for it until later.

Last week IBM announced the z10 Business Class, a smaller counterpart to its big honcho, the z10 Enterprise Class. The z10 BC starts at about $100,000. It’s far less than a seven- or eight-figure EC, but still, it’s not pocket change in this current economic climate. As part of the announcement, IBM is offering a financing deal. Order one now, and you don’t have to make a payment for 90 days.

Can you see the commercial now? Can you see the different color balloons floating around the mainframe retail store, luring unknowing customers into buying a mainframe? Actually, though, financing deals in the IT world are nothing new, said Mike Kahn of The Clipper Group.

“It’s an incentive to bring business in this year,” he said. “They’re ramping up product and getting it ready to ship. A lot of people would say they don’t have any budget left. The economy is down and they’ve blown their budget for this quarter. They might not have any money this quarter but they might have it next quarter.”

Kahn added that IT vendors often have end-of-the-year deals, where they might throw in extra memory at no cost to try to get product out the door by the end of the quarter. IBM’s move is just another end-of-year deal.

Specialty engines cut in half

The other big financial incentive that IBM announced at the same time was that the specialty engines for the z10 BC would be half price — so about $45,000 instead of almost $100,000.

The specialty engines include the zIIP, zAAP and IFL, which are geared toward running database, Java and Linux applications, respectively. Those less fond of IBM and its mainframe have called them stripped-down z/OS engines, and to an extent, that’s true. Either way, the specialty engines have offered a way to get traditionally non-mainframe applications — Linux and Java especially — running on big iron in a consolidated fashion on top of z/VM, big iron’s virtualization operating system.

During the announcement, IBM officials said they are gearing the z10 BC to be a big consolidation play, where users take a bunch of their x86 Linux servers and stuff them onto the mainframe. Reducing the cost of the specialty engines could help it work.

 ”They believe this is a real opportunity for them and customers to rethink what they’re doing with these more open applications,” Kahn said.

Kahn added that if you’re running a z9 Business Class with specialty engines, and you decide to upgrade to the z10 Business Class, IBM will give you the upgraded specialty engine at no charge.

“That’s like if somebody says they’ll take old used tires off my car and give me better tires,” he said. “I could use a few of those deals in my life.”

Cloud computing = mainframe rehash?

Andi Baritchi, who is listed as being a senior security consultant for a Fortune 100 company, recently wrote that cloud computing is the completion of a loop back to the mainframe.

How so? It takes the computing power back out of the hands of consumers and puts it back into the hands of the internal IT employees and third-party vendors. Take a look:

Since history does tend to repeat itself cyclically, I see this whole cloud computing movement as nothing more than a reincarnation of the classic mainframe client-server model. People want painless access to their data and applications from wherever they are, from whatever electronic gizmo they happen to be using. Sometimes we’re on a computer. A laptop. A desktop. A smartphone. A smartTV. A smartfridge. A smartcar…?

There has been a lot of talk about cloud computing being like utility computing, or grid computing, or other types of recent client-based computing infrastructures. Baritchi is saying here that it goes back even further, to the days of dumb terminals and mainframes.

According to Baritchi, the vendors out there that are doing the best job of it are Apple with the iPhone, and Google with Android. Either way, he’s excited about the reversion:

As a side note, this is a very exciting time for me being a security and privacy guy. This feverish movement back to client server to the cloud will see many security lapses. It’s gonna be a fun ride.

Mainframe revenue the brightest star among IBM systems

On the hardware side, mainframe revenue for the fiscal third quarter performed far and away better than other IBM server platforms, presumably because Big Blue just came out with the System z10 earlier this year.

System z revenue was up 25%. Meanwhile, the “converged System p” platform, which is the System i and p combined, rose 7%, although that increase is somewhat misleading because it compares to only the System p platform from last year. System x revenue plummeted 18%.

IBM added that there was “double-digit growth in all geographies” for mainframe server revenue. Total delivery of mainframe computing power, measured in millions of instructions per second (MIPS), increased 49%.

Major System z announcement next Tuesday

IBM will hold a System-z related announcement at 10am Eastern next Tuesday, Oct. 21. You can register for the System z webcast here.

The event will include Karl Freund, VP of System z marketing, and Mike Augustine, the System z product manager, as the featured speakers. Others involved in a live Q&A at the event will be:

• John Eells, z/OS Technical Marketing
• Allen Marin, System Storage Market Manager
• Nancy Scala, Linux and z/VM Product Manager
• Kurt Johnson, System z Software Market Manager
• Greg Hutchison, IBM System z, Certified IT Specialist and Advanced Technical Support

Here’s a summary of the event:

There is new System z technology on the horizon that could change the way your organization thinks about mainframes. Technology that delivers the granular scalability, flexibility, and resiliency you need — at the lower capacity entry point you want. Its advanced technology fights old myths and perceptions – it’s not just for banks and insurance companies.

For any organization that wants to ramp up innovation, boost efficiencies and lower costs – pretty much any enterprise, any size, any location, z Can Do IT.

The Webcast description repeats “z Can Do IT” enough times that I’m wondering whether Rob Schneider might make a guest appearance.

Anyway, my guess is that the announcement will be for the “Business Class” version of the System z10 — a smaller version of the big boy. But I don’t have any definitive knowledge on it.

The Mainframe blog on Typepad had a blurb about this but for some reason has taken it down.

DoD mainframe growth at 10% a year

John Garing, the CIO and director of strategic planning for the Defense Information Systems Agency, reveals in an interview that the Department of Defense is increasing its mainframe spending by 10% a year.

The Q&A by The Wall Street Journal touches on a host of issues, including how the DoD has cut its data center spending 25% by consolidating facilities and only paying for the processing power it needs. Here’s what Garing had to say about the department’s mainframe use:

First of all, mainframe security is very good. Second, they are very fast and powerful machines. Third, they can do things that it would take racks and racks of servers to do. So while there aren’t very many new applications being written for the mainframe, we’re still finding new ways to use them.

An example is our payrolls, which already ran on a mainframe. We [wrote software to let people access their payroll information] so we just did it for the mainframe. We’re buying new mainframes [when the old ones wear out]. There’s no plan to go wholesale to client server like there was years ago.

The rest of the interview, though unrelated to mainframes, is still interesting.

A Q&A on mainframe application modernization

When you think of mainframe application modernization, you tend to think of two things: SOA (Web services) or migration. That is, put a pretty face on it or get off. Well, Tim Pacileo, the principal consultant at Compass America Inc., said it doesn’t have to be that way. Compass is an IT and business operations consultant based in Naperville, Ill., just outside of Chicago. Pacileo sat down with me to talk about application modernization on the mainframe. There’s no doubt that a lot of the talk is about Web services, but that’s not the only option.

Compass just came out with five tips to build your application modernization case. Talk about tip 1 — justifying the cost — and how it relates to mainframes.

On the mainframe, when a particular application runs out of capacity or fails to meet the needs of the business, rather than touch that application and go through several reversions and updates, most organizations will just throw people at it. On the other end, we’re seeing a lot of core legacy applications with a lack of flexibility preventing the environment from doing what it needs to do. We’ll look at various options, such as replacing part of the core application or migrating off the platform.

Is it always about getting rid of the application or migrating?

There are some options. One company I worked with a few years ago, their goal was to not get off the mainframe for another 10 years. There was no compelling business driver to get off the mainframe. What we looked at there was enhancing that application with some Web-based solutions so people could get in with a browser and look at their data.

Are migration and browser-like functionality the only options?

No. You can change the nature of the mainframe. With another customer, we are moving them from Cobol CICS to turning the mainframe basically into a large database server and enhancing that with DB2 applications.

We’ll see some customers come to us and say that they want to stay on the mainframe. It’s a proven environment. I can have hundreds of applications on a mainframe environment. We’re talkinga bout virtualization now, but the mainframe has been virtualized for decades. And we do see organizations running multiple Linux applications on the mainframe. We’re seeing some growth in that area.

Others say that they want to get off the mainframe because they don’t want to be the last person standing. Some of those core applications are too expensive to modernize or migrate, and so they’re looking to move to some off-the-shelf applications.

What are some of those core applications?

Normally they’re homegrown applications. Health care is one. Government has a lot of homegrown applications, and they don’t have the functionality they need. The legislature keeps making changes, and it’s a Herculean effort to keep up with it.

Mainframe gives some Iowans a tax break, but this isn’t a good thing

Linn County, Iowa residents will likely get about a one-month reprieve on paying their property taxes, thanks to the mainframe.

But this isn’t a good story.

The flood earlier this year that devastated many parts of the state also damaged a mainframe that the county runs many of its tax applications on. Normally, property tax bills are mailed in August and due by the end of September. Now the county treasurer, Mike Stevenson, “anticipates” that they’ll be mailed in September, and residents will then have 30 days to pay them.

Perhaps the county should think about deploying a parallel sysplex, or at least devise a more-robust disaster recovery plan?