For horizontal technologies, like integration, it is difficult to quantify Return On Investment (ROI) because it underpins many business systems. What can be quantified is the Total Cost of Ownership (TCO). Nowhere is TCO more important than mainframes, where a single IBM System z10 can have the capacity of 1500 or more Intel servers. Licensing hardware and software on such an enterprise scale can be costly, so there is, and always has been, a need to manage this capacity and understand how its resources, such as processor time, are being allocated.
IBM’s line of specialty engines, including the System z Integrated Information Processor (zIIP), are designed to help lower mainframe TCO by processing qualified workloads rather than having this work processed on the General Purpose Processor (GPP). These engines are just like a GPP except:
- Their capacity is typically not used in calculating software licensing fees based on mainframe capacity.
- Their processing speed is not governed, that is they run at full speed all the time
- Their processing capacity is enormous – a single zIIP engine for an IBM System z10 machine has a capacity of 920 MIPS.
However, IT decision makers requested we provide more estimates of the potential savings based on THEIR workloads. In response we partnered with Wintergreen research (www.wintergreenroi.com), a well-respected analyst firm specializing in TCO/ROI analysis, to deliver a Web-based calculator (www.datadirect.com/products/mainframe-integration/shadow-rte/shadow-tco-home/tco-calculator/index.ssp). The Calculator models the potential capacity savings, measured in MIPS, as well as the monetary savings. It uses what is called the Willhoit constant, named after Gregg Willhoit, our Chief Software Architect, who developed the algorithm that estimates the offload of DataDirect Shadow processing. Today the calculator covers two processor-intensive types of integration processing: Web Services and SQL. In as little as an hour, our field-engineering team can quantify the savings using your workload profile for:
- Number and type of Web services (requester or provider) or SQL statement type (join, aggregate, etc.)
- Size of SOAP payload or estimated result set size for SQL
- Invocations over a modelled timeframe – such as per-day or per-peak period to help model peak capacity requirements
- Cost per MIPS. Because cost can be calculated differently, the model offers the option to use the comprehensive Wintergreen mainframe-costing model, which includes hardware, software, data centre, and labour costs, or a way to use your own numbers.
In as little as an hour, IT can be in a much stronger position to provide detailed and accurate information to support ROI analysis of not only mainframe integration investments but the potentially large MIPS dividend available to the entire mainframe from utilizing zIIP specialty engines to process up to 99% of the integration processing performed by Progress DataDirect Shadow.
Thanks Jeff for being our first guest blogger. And remember, there's still time to complete the mainframe user survey or a vendor entry for the Arcati Mainframe Yearbook 2010.