Performance and Workload Management: Still “Table Stakes” in Mainframe Management
The mainframe needs to perform; yet cost pressure is everywhere.
By Jeff Albright, GTSG12/18/2017
The mainframe needs to perform; yet cost pressure is everywhere. With additional budget at a premium, “throwing hardware” at a perceived capacity problem is no longer a solution—indeed, it never was. Increased capacity costs cascade through the ISV stack.
The now-scarce skilled resources are stretched. When a budget to replace headcount exists, funding is rarely available to shadow before retirement. As a result, vital tribal knowledge is lost. This combination of busyness and lost experience has resulted in abandoning the once-common performance and workload management disciplines that helped get the most out of mainframe investment.
Performance management takes a back seat until the phone starts ringing. That’s the “sneak attack,” with SLAs increasingly missed; help desk lines ringing; and availability questioned: is it “available” if it doesn’t meet the needs of the business? Seasonal pressures can spotlight smaller non-seasonal issues.
Workload management makes resources available to who needs it–not who wants it. There’s nothing wrong with a processor running 100% if SLAs are met. Capacity planning may become excited when processors reach a threshold, but finance appreciates the maximum use of an asset. We understand managing the rolling four-hour average as the end goal, but it’s not the starting point for the engagement.
Performance management equals MIPS reduction, plain and simple. The focus is tactical, and task-oriented; on understanding pain points and focused on pressing issues for the creation of working room. Scope includes infrastructure, applications, scheduling-whatever leads to an issue.
One recent engagement illustrates this well: at year end, an outsourcer pressed their client to upgrade an engine by year-end. The client declined and undertook a performance & workload engagement. Utilizing the approach described here, they found enough headroom that nearly two years later, they were operating with the original engine and had avoided the upgrade, and the associated expenditure.
The measure of success is simple: when you’re performing the same (or more) work with less computing resource. In today’s cost-challenged environment, a performance and workload program should be “table stakes” for every mainframe shop.
From his start in 1978 as an assembler programmer to his current role as Founder and Managing Partner of Global Technology Solutions Group, Inc., Jeff Albright's company now supports a meaningful portion of the 40 million MIPS in production worldwide.