Initial announcements include infusing CA's array of mainframe software with electronic software delivery (enabling utility-driven installation); the arrival of CA Mainframe Software Manager (MSM), aimed at automating installation and maintenance; best practices guides, interoperability certification (with other CA mainframe products) and integration with the IBM Health Checker for z/OS framework (utilities checking the effectiveness of operational CA mainframe software); and the Mainframe Value Program and Mainframe Solution Center for customer support. CA has also announced CA Compliance Manager for z/OS security/compliance policy management, the first new product that is MSM-compliant out of the box.
What effect should users expect Mainframe 2.0 to have on costs? And how can it be applied most effectively?
A mainframe-driven TCO strategy
The best way to view Mainframe 2.0 is in the context of overall IT cost cutting that includes the mainframe. Over the last few years, mainframe total cost of ownership (TCO) studies have shown that using virtualization software to consolidate more than 20 Unix, Linux or z/OS (or migratable) applications on a mainframe reduces the following costs:
- Software license costs, both because software at all levels may charge per system rather than per virtual machine, and because the mainframe's zero-cost specialty processors sometimes mean that software can scale higher without needing new licenses;
- Administrative costs (including upgrade costs), because administering one big system is cheaper than administering more than 20 smaller systems;
- Power/cooling costs, both because of the mainframe's design and because using one system avoids the added heat generated by additional redundant components (such as power supplies) in multiple systems.
These reduced costs tend to counterbalance higher up-front hardware costs, and the payback period, beyond which the mainframe is less expensive than multiple non-mainframe systems, typically ranges between six months and 1.5 years. Moreover, long-term trends are that each of these costs is an increasing percentage of overall TCO, so that the mainframe's TCO advantage is very likely to increase over the next three years. Initial studies suggest that the mainframe's three-year TCO advantage versus Unix/Linux systems can range from 10% to 40%, and the greater the number of applications that are virtualized or consolidated, the greater the advantage.
Therefore, an effective user strategy for reducing IT data center costs by double digits is consolidating applications in virtual machines on the mainframe. Additional power cost savings may come from data center redesign and use of mainframe software to monitor energy usage. But the main follow-on mainframe cost reductions lie in the area of administrative software: in particular, in reducing the people costs of mainframe administration.
Thus, CA's Mainframe 2.0, with its primary focus on automating CA mainframe software installation/upgrade and mainframe administration, will have a major effect on z/OS application and system, database, and storage administration. Because the CA products designed for these areas are valued for sophisticated solutions to complex problems, it's likely that the larger the system, the greater the percentage of administrative savings.
Mainframe 2.0 impact on administration costs
Although there is no definitive data, my TCO studies have led me to believe that today, database administration involves the most "people cost" per VM, with application and system administration coming in second, and storage administration typically involving the least administrator time. Software deployment/upgrade/license management is not typically a major people cost, but is risky -- problems can mean not only a large amount of administrative costs and time, but also a major impact on system availability.
Mainframe 2.0 has a significant impact in all of these areas of administrative cost. The most obvious impact is in the area of deployment and upgrade, with electronic software delivery and MSM. However, in the typical case this will have less impact than the automation improvements in CA's mainframe database administration and utility/system administration products.
The maximum impact from Mainframe 2.0 may well be in database administration costs. What is the "outer bound" of this impact? My studies of small and medium-sized businesses suggest that a zero-administration database can yield administrative costs that are one-tenth as much as an Oracle database. There is only so much that automating the administration of performance-tuned, high-scaling mainframe databases can do to reduce this gap, but it appears from recent vendor efforts to automate enterprise database administration that effective automation can cut administrative costs in half. For VMs running CA IDMS or CA Datacom, this would mean administrative cost reductions of at least 20%, and therefore 5 to 10% overall TCO reductions. In the more frequent case where CA tools supplement basic DB2, IMS or VSAM administration, the overall effect might be 2 to 5%.
Moreover, there is a side benefit of increasing mainframe administration automation: less dependence on the baby boomer administrative personnel that are now finally beginning to retire. CA Mainframe 2.0's effects will be not only on TCO, but also on the longer-term risks of dependency on these administrators to keep the business running.
When an IT department needs to cut costs now, not next year, the ability of CA Mainframe 2.0 to deliver immediate administrative cost savings to supplement a mainframe consolidation strategy should put it high on IT's priority list. And the biggest bang for the buck in using CA Mainframe 2.0 may well be in database administration.
ABOUT THE AUTHOR: Wayne Kernochan is president of Infostructure Associates, an affiliate of Valley View Ventures. Infostructure Associates aims to provide thought leadership and sound advice to vendors and users of information technology. This document is the result of Infostructure Associates-sponsored research. Infostructure Associates believes that its findings are objective and represent the best analysis available at the time of publication.
What did you think of this feature? Write to SearchDataCenter.com's Matt Stansberry about your data center concerns at firstname.lastname@example.org.
This was first published in June 2009