Over the last year, IBM has made a major push in both "green" hardware and software, and the mainframe has been
a big player. The z10 is taking a leadership role by supporting data center energy use monitoring, using virtualization to pack more computing power into a single machine, and figuring out how to reduce per-chip and per-disk energy requirements. According to IBM, the mainframe consumes 80-90% less energy than an equivalent-computing-power PC-server distributed system.
According to a recent UN-sponsored scientific consensus, the effects of global warming are real and already serious. In order to reach a "steady state" that will minimize any additional warming beyond what's already "baked in," the world needs to at least slightly decrease overall net carbon emissions each year from now on, by reducing energy usage and/or reducing the amount of carbon emitted by a given amount of energy usage.
IT is already up to approximately 1% of the world's energy usage (one IBM estimate has data centers using 183 billion kilowatt hours of electricity in 2007), an astonishing rate of growth of energy usage to reverse. Meanwhile, by some projections, global storage size has been increasing and will continue to increase by 60% year-to-year over the next five years, which historically has meant a similar yearly increase in server and overall IT energy consumption, and therefore, given the way electricity is now produced, in IT-driven carbon emissions.
What role can the mainframe play in solving the climate problem?
Clearly, the mainframe-focused data center redesign mentioned above can make an astonishing dent in data-center consumption; but it isn't enough. According to my crude computations, in order to counteract the effects of storage growth within a single data center, mainframes need to reduce their energy usage by about 40% per year, year after year. In other words, existing improvements are only enough to deliver single-data-center energy reductions for a little over three years; after that, further "design breakthroughs" are needed, even though the easy redesign targets have already been identified. Improvements in the mainframe's virtualization software, which would allow more computing with the same energy consumption, appear likely to be incremental. Simply reducing energy consumption per data center isn't enough if an organization proliferates data centers or fails to rein in the energy consumption increases from laptops and PCs. Additionally, mainframe redesign can do little about the 99% of energy consumption that is not attributable to IT.
I would argue that the mainframe can indeed play a major role, not only in the data center but also across the organization, in achieving "steady state" or lower energy consumption. Specifically, the mainframe is a good central location for monitoring not only IT, but business processes in general, for their energy consumption implications. This means that the mainframe should be the primary site for, one, monitoring/metering/improving data-center energy usage; two, monitoring/improving energy usage of organizational computing resources outside the data center; and, three, measuring/monitoring/improving energy usage of organizational business processes.
The case for the mainframe as a data-center "green monitor" is a slam-dunk: it's there, it has a lot of ability to improve energy usage on its own, and IBM has begun to offer extensive monitoring/metering/fine-tuning mainframe software tools for this purpose.
Energy monitoring can center on the mainframe
Monitoring computing energy usage outside the data centers (and coordinating across data centers) isn't as easy. However, frequently the mainframe is the locus of asset management software that profiles IT assets across the company. It is not a big step for any organization to use this data to develop an informal picture of computing energy usage, and to invoke vendor services to advise on improving energy usage by redesigning the allocation of computing resources to end users. Remote energy-monitoring software would be nice; but it isn't needed to make a serious dent in the problem.
But the real payoff for any organization is in applying "green monitoring" to business processes, and this is where IBM and other software/hardware vendors have really fallen short up to now. Granted, IBM is offering travel- and paper-reducing software, including repurposing Lotus Notes collaboration solutions to avoid employee travel and advocating increased use of Content Manager and the like for saving paper. However, both BPM (business process management) solutions and enterprise applications such as ERP (enterprise resource planning) from suppliers such as Oracle and SAP typically do not incorporate energy consumption considerations. Enterprises will simply have to piggyback on the mainframe's documentation of those business processes, with some help from business and IT consultants.
So the mainframe can play a significant role now in IT "green monitoring"; and when business-process "green monitoring" arrives, the mainframe can play a vital role in an organization's overall green efforts. Meanwhile, mainframe users should bear in mind that we are not at the end of our green efforts, nor even at the end of the beginning of these efforts; we are at the beginning of the beginning, and planning should begin for a very long-term effort.
ABOUT THE AUTHOR: Wayne Kernochan is president of Infostructure Associates, an affiliate of Valley View Ventures.