Wednesday, February 18, 2015

Mainframe's High MLC Cost - Can It Be Reduced?

by Brenda J. Christie and Eloy Cruz-Bizet

Mainframe Computer Monthly License Costs Keep Rising
Rising MLC Costs

There has been a lot of speculation as to whether IBM's billion dollar bet on the System z13 will pay off ("Will Big Blue's billion-dollar bet on its new Mainframe pay off", Forbes article "Next Week's Bloodbath at IBM Won't Fix the Real Problem.")  Some companies, like Citigroup, as reported in American Banker's January 14, 2014 article, believe the bet will pay off, have already bitten the hook and availed themselves of the System z13.  I, for one, believe IBM has a chance, but its fortunes will not go unchallenged.

Not Hearing Voice of the Customer?

As alluded to in the Forbes article, IBM does not appear to be addressing its customers' needs, especially where software licensing is of concern. One great concern in IT management is cost reduction.  As reported in BMC's 2014 Annual Mainframe Research Results, the top mainframe priority for the upcoming year was to lower MLC.  Indeed, at the time of this writing there were 1,500 YouTube videos surrounding the topic of reducing MLC.  Google search on how to reduce mainframe MLC returned 21,700 results.  How to reduce mainframe MLC costs seems to be on a lot of people's minds.

What Is MLC and Why Does It Matter?

Mainframe-related MLC, or Monthly License Costs, is defined by IBM as "Monthly License Charge (MLC) metrics are those where a recurring charge applies each month. This charge includes the right to use the product, and also access to IBM product support during the support period." These are usage costs for system, subsystem and transactional software including the z13 operating  system itself and can amount to millions of dollars annually.  In a 2013 Forrester artlcle, "Get A Grip on those IBM mainframe MLC Costs" estimates for MCL were 30% or more of software budgets.
This is a huge cost being diverted from modernization and development in new business.  Despite Forrester's admonition in 2013, MLC remained an issue at the end of 2014, giving credence to the supposition that IBM is not adequately addressing its customers' dissatisfaction.

Partitioning LPARS to Reduce Cost or Let Them Eat Cake

That reluctance to address customers' concerns regarding high MLC costs may, in part be due to hubris, but may also be attributable to IBM's own answer on reducing MLC as noted in its 2006 coursework guide, "Introduction to the New Mainframe Chapter 2 Hardware Systems and LPARS," in which it recommends partitioning the box into LPARS, or Logical Partition.  An LPAR is similar to a mainframe within a mainframe complete with its own storage, processor and storage.  As MLC can be, in some instances, applied to the box itself, partitioning the box into several independent mainframes would allow for different applications such as CICS, DB2, Linux to run separately and negate the cost one would see if 3 different mainframes were maintained as well as the 3 separate MLC costs.  While partitioning as a strategy does limit the MLC growth, it does not reduce it. Companies who are being pressured to reduce costs cannot show that reduction even with partitioning.

To its credit, IBM did publish a white paper, Optimize MLC Expenses of Your Mainframe Data Center with System Automation for z/OS (see below), and in June 2014,  announced support for BMC's Cost Analyzer for zEnterprise(R) version 1.1, a graphical product which analyzes SMF/RMF cost and utilization data to calculate MLC costs by CPC while identifying workload peaks on LPARs and CPCs.

Is It In IBM's Interest to Reduce MLC?

One must also be cognizant of the fact that, as noted by Steve Mills, Executive Vice President of IBM Software and Systems, during a Credit Suisse Group AG Technology Conference last month, the mainframe software business is a multi-billion dollar software business on the mainframe.  That being said, it is probably not in IBM's interest to sell less software, or eliminate limitations such as the one requiring DB2, CICS, IMS to be on the same LPAR. Factor in the slow growth in mainframe sales, and the incentive to keep such requirements becomes even stronger.  It is perhaps because of these factors that IBM has not actively gone out of its way to address its customers' needs to reduce expenses through MLC reduction.

Someone Else was Listening to IBM's Customers

BMC seemingly took the bull by the horn and went a step further in September 2014 when it announced its very own software solution for reducing MLC - the Subsystem Optimizer for zEnterprise.  Also known as Subzero, the software solution is purported to reduce monthly license costs by as much as 20% or more. Extending the analytical capabilities of the Cost Analyzer, Subsystem Optimizer allows customers to redirect work between different LPARS, somewhat similar to an air traffic controller.  BMC's Subzero software breaks the dependency to have CICS, DB2 and IMS within one LPAR, necessitating an additional, separate LPAR for other applications. The inherent problem caused by requiring CICS, DB2, IMS to reside on a single LPAR is the potential under-utilization of other LPARs within the subsystem.   The information needed to process requests between CICS, DB2 or IMS software residing on different LPARs is handled through BMC Subzero APIs.  Being able to separate DB2, CICS, IMS onto separate LPARs is a valuable feature as MLC is calculated based on overall usage such that an under-utilized LPAR is charged in excess of its actual usage.  By enabling the redistribution of workloads to under-utilized LPARs ROI on the heretofore under-utilized LPAR is improved.  Subzero's flexibility in redirecting workloads in also of value in Business Continuity and Disaster Planning scenarios, further justifying its cost.

Customers May Prevail After All

To the extent that BMC is successful, customers may finally have a way of reducing MLC cost and using the savings to expand into areas such as Cloud Computing and Mobility. I had the opportunity to ask Ann Duhon, of BMC questions regarding Subsystem Optimizer's success to date to which she replied " ."


IBM's System z13 has a lot of bells and whistles, as noted in my earlier post, "15 Reasons the IBM System z13 Just Might Succeed."  Also noted in my "Is the Buzz Surrounding IBM's System z13 Just Hype?" post are some hidden costs in moving to the z13.  Given that IT Management's mandate is to chisel away some of the thousands, if not millions of dollars spent annually on mainframe Monthly License Costs, I think it is more likely that companies will elect the path that meets that mandate, that is to say, they will chose to try the BMC Subsystem Optimizer.

As always, time will tell.

Bye for now,

Brenda J. Christie 


Optimize MLC Expenses of Mainframe Data Center with System Automation