Last week I spoke with Chris O’Malley, the head of CA’s mainframe business unit and a keynote speaker at the upcoming Share conference in Denver, about some of the challenges facing mainframers today. Jim Hood, a mainframer at Siemens Healthcare, wrote to me and had this to say in reply:
“Cost, I think, is primarily at the heart of the mainframe dilemma: hardware, software and people. Skills, or the concern of those disappearing over time, is secondary. Hardware and software costs, and the perception that these are so much higher than a compatible server based operation, helps fuel the urge to displace this platform. I still cannot be convinced that the TCO is significantly higher than that of running thousands of servers when you add license costs, all the replaceable components and ongoing support (manpower) for upgrades, maintenance, asset management, etc. Even so the movement continues.
The related costs for the skill sets needed to run a mainframe operation is the other fly in the ointment. Older mainframe programmers/analysts get a bad rap from a cost standpoint even though their long tenure at many organizations has made them successful (the business and themselves) while their long, loyal labor yielded advancement and increased salaries. That isn’t a good mix in today’s marketplace where low costs rule. I wonder what happens when the same scenario plays out 20 years from now when the current younger, lower cost workers are not so young or low cost anymore.
I’m not sure if there are any real replacement plans from a talent standpoint. That would mean acquiring “young” labor while keeping “older” labor which means increasing costs even more. In today’s world cost reductions don’t have the luxury of time on their side — immediacy rules — so while there might be a concern about keeping talent the reality, I think, largely depends upon a balance sheet. Even for those who could stay (to mentor) many mainframers have already been in the business 30-40 years. It is going to take a lot of activity to get the twentysomethings thinking that zOS is cool. By the time that happens (if ever) there won’t be many mentors left — whether by choice or not.”
BMC acquires MQSoftware
This week, BMC software acquired message queuing software firm, MQSoftware. BMC is touting the acquisition as a boon for mainframe shops that need to manage the middleware for applications that span both distributed and mainframe platforms.
For certain kinds of big applications, message queuing software is a necessary glue for distributed applications because sometimes servers are down or systems are just running slowly, and you don’t want a transaction to stall just because a backend system is not performing well. So instead of hard-coding the transactions pieces to not continue unless each stage of the transaction is completed, you carve up the transaction into messages and then queue them up on different parts of the application’s servers, explained Timothy Prickett Morgan in his writeup of the BMC-MQSoftware news.
CA adds Eclipse support to CICS application testing
In other mainframe systems management news, CA added a new GUI based on Eclipse for its testing and debugging tools, CA InterTest Batch and CA InterTest for CICS. CA said this new GUI will make it easier for novice mainframe staff to execute core testing, and debugging tasks. CA also released a new version of CA SymDump for CICS, its advanced abend analysis solution.
On July 28, IBM announced it would acquire business intelligence software company, SPSS. The merger may have a big impact on the mainframe’s business intelligence abilities in the future.
IBM has done a lot of really solid work making the mainframe less expensive for non-CICS and IMS workloads like Linux (IFL), DB2 (zIP) or WebSphere (zAAP). IBM is determined to drive datawarehousing workloads to the mainframe. But SAS Institute was a “stick in the mud”, effectively forcing users to pay capacity-based mainframe charges, and so making it less likely customers would run Big Data analytics on z. Well now IBM is in a great position to offer specialist offload processors for data analytics workloads, but also push SAS Institute into a price war that can only benefit customers interested in mainframe consolidation- and don’t think that’s an isolated group.
Cisco XRC Acceleration, a WAN acceleration tool Jointly developed by Cisco and IBM was designed for customers deploying IBM’s mainframe disaster recovery offering, z/OS Global Mirror. The Cisco XRC Acceleraton tool accelerates data traffic traveling very long distances over the wide-area network (WAN) reducing bandwidth consumption and shrinking update windows.
There are a slew of other Cisco announcements attached to the press release.
Timothy Prickett Morgan apparently got his hands on a Novell presentation about SUSE Linux Enterprise Server (SLES), its iteration of Linux that is the most popular distribution running on the mainframe on top of z/VM.
Morgan reports that IBM and Novell are close to announcing a limited time slicing and dicing of support license costs for SLES, to the tune of as much as 50%. Here are the details:
The basic license cost for SLES 11 (and any SLES release for that matter) on IBM mainframes is $11,999 per engine for a one-year contract. That gets you installation support and Web support thereafter. A standard 9×5 business hour contract with human beings providing support for a year costs $15,000 per engine, and for premium 24×7 support, it costs $18,000. With this impending promotion, basic SLES 11 support prices per mainframe engine are being cut 40 per cent to $7,199, standard support is being cut 32 per cent to $10,200, and premium support is being cut 27 per cent to $13,200.
There is also another, seemingly separate promotion (it doesn’t look like the two can be combined):
If mainframe shops want to save some additional bucks after that, they can get a three-year support contract at 35 per cent of list and a five-year contract at 47 per cent off list. This can be some pretty substantial savings. A five-year SLES 11 premium support contract runs $44,999 per engine after the discount, which works out to $9,000 per engine. This promotion will be offered on the midrange System z10 BC mainframes (that’s short for Business Class, and offering from 1 to 5 mainframe engines), not the full-blown z10 EC machines (short for Enterprise Class, and spanning from 12 to 64 engines). This promotion will run until December 31, apparently. It is not clear when it will be launched, but probably this week or next.
In the presentation, Novell claims that it has 1,300 customers running Linux on 4,000 Integrated Facilities for Linux (IFLs), which is the mainframe specialty engine built to run Linux.
OK, here we go. Many were wondering what IBM’s reaction to the Neon zPrime software would be. As you may know, the purpose of zPrime is to offload work from a mainframe’s central processors to specialty engines such as the zIIP and zAAP, saving on software costs and possibly delaying a multimillion-dollar mainframe upgrade. What Neon claims is that zPrime can offload a lot more than just DB2 and Java workloads, which is what the zIIP and zAAP are aimed at, respectively. One beta customer said they could save $10 million a year using zPrime.
Initially, IBM had no response, saying it needed to look into the product first. Neon was also being coy, not releasing any pricing on their product even though it is supposedly in general availability.
Now IBM is warning its mainframe customers.
In a letter dated July 10 and addressed to a generic client, IBM restates that it is trying to learn more information about the product “and how it is supposed to work, but the vendor has not yet offered to provide IBM with any detailed information and seems to be restricting access to such information.” Though I’m not positive about what “restricting information” means, I heard from a reliable source that a Neon Webcast last week on zPrime wasn’t allowing IBMers to register. The letter continues:
In general, any product which is designed to cause additional workloads, not designated by IBM or other (software) providers as eligible to run on the Specialty Engines, to nevertheless to be routed to a Specialty Engine should be evaluated to determine whether installation and use of such a product would violate, among other things, the IBM Customer Agreement (for instance, Section 4 regarding authorized use of IBM program products such as z/OS) and/or the license governing use of the IBM “Licensed Internal Code” (frequently referred to as “LIC”) running on IBM System z servers, or license agreements with any third party software providers.
IBM would also caution its customers regarding any claimed ability to reduce IBM Program license charges by off-loading workloads to Specialty Engines beyond the eligible workload identified by IBM. IBM’s applicable pricing terms governing Eligible Workloads on zIIPs and zAAPs will not apply to zIIPs and zAAPs running anything other than IBM specified eligible workloads. Therefore, customers should not anticipate any reduction (and may actually experience an increase) in the IBM Program License Charges associated with non-Eligible Workloads which may be off-loaded to IBM Specialty Engines, since the non-Eligible Workload running on the Specialty Engine will cause the software running on the Specialty Engine to be chargeable. IBM cannot comment on the potential impact on the software charges from other third party software providers.
So what will happen next? Will IBM still try to work with Neon to make this product a viable one for reducing mainframe costs? Will the negotiations turn contentious, possibly resulting in lawsuits and/or an acquisition, as was the case with Platform Solutions Inc. (PSI)? Rich Ptak, an analyst with Ptak, Noel & Associates, gave his opinion on the two sides of the coin.
“It could provide [IBM with] the significant opportunity to reinforce the message that the mainframe can be a very attractive workload platform,” he said. “It could really go to the heart of the challenge today that HP and Microsoft and distributed folks make about moving off the mainframe. They could also just as easily decide that they’ll just start including MIPS [million instructions per second] on the specialty engines in their pricing. To me, in many ways, that would be self-defeating.”
I discovered this IBM letter through DataDirect, another mainframe software vendor that sells a product for offloading work to the zIIP. Neither IBM nor Neon Software got back to me yesterday for an interview.
Now, I’m not here to recommend any organization’s software – they pay people to do that, and I haven’t noticed a brown envelope full of bank notes being slipped into my hands – but it seems this kind of software is going to make NEON popular with everyone except IBM! Obvously, MSUs (Metered Software Units) are IBM’s way of ensuring a regular income – and it’s based on GPP usage. If people find a way of running their software on zIIP and zAAP specialty engines then IBM will have to come up with a new pricing model.
It seems that they best policy for users is to get in quick and pay for zPrime with the savings that will be made on monthly licence fees. And then bank all future savings until IBM changes the pricing rules.
IBM announced its second-quarter financial results yesterday. Mainframe revenue tanked compared to the same quarter last year, dropping 39 percent.
Other platforms did poorly as well. System x decreased 22 percent and the Power-based servers, which IBM calls “converged System p,” decreased 13 percent. System revenues overall decreased 26 percent.
The Mainframe Typepad blog says System z mainframe revenues were expected to be down simply because it’s being compared to the second quarter of last year, which was the first full quarter that the new System z10 mainframe was being sold.
The decrease is also just part of an overall slow economy. HP’s second-quarter results were similarly bad, with its servers and storage division down 28 percent. Dell hasn’t released its second-quarter results, but its first-quarter results saw a 29 percent drop in servers and networking.
According to the Palestine Herald, thieves recently stole a mainframe computer from the Trinity Valley Community College campus. The mainframe contained students’ sensitive personal information. Anderson County Sheriff Greg Taylor said “Of course, they cut a lot of wires (to take the mainframe computer). There will be a lot of damage.”
IBM announced yesterday that z/VM 6.1, the newest version of its mainframe virtualization operating system, will be out by year’s end.
The z/VM operating system has become more and more relevant over the past few years as companies have taken advantage of the virtualization appeal of the mainframe to run hundreds, or even thousands, of virtual Linux servers on big iron. There have been estimates that more than half of new mainframe MIPS are on zLinux.
The last version of z/VM, version 5.4, was released in June 2007, so this new version has been a while coming compared to the operating system’s previous release schedule.
The new version, 6.1, will be the first built on a new architectural level set (ALS), which is a fancy way of saying it will only run on the System z10 mainframes and future models. The expected end-of-life IBM support date for z/VM 5.4 is September 2013, so those running z9 mainframes and earlier have a few years to upgrade, find support from a third party, or go without support.
Some highlights from the new release:
- Guest LAN and Virtual Switch (VSWITCH) exploitation of the Prefetch Data instruction to use new IBM System z10 server cache prefetch capabilities to help improve the performance of guest-to-guest streaming network workloads
- Closer integration with IBM Systems Director by shipping the Manageability Access Point Agent for z/VM to help simplify installation of the agent
- Inclusion of post-z/VM V5.4 enhancements delivered in the IBM service stream.
Perhaps just as importantly, IBM says that z/VM 6.1 provides a formation for future z/VM enhancements, including the ability to allow multiple z/VM systems to act as one server image, and to allow live guest relocation from one system to another.