While traveling on business, I heard a refrain whose time I thought had passed in referenced to System z. No, it was not the one that claims the platform is ‘more expensive’. (There are reams of material and experience to lay that urban myth to rest!) This was a comment bemoaning the lack of new workloads for System z. What??!
Perhaps the biggest door opener for new workloads have been the capabilities to bring Linux workloads on board with the enabling capabilities of VM and IFL engines on System z. Not only does this strategy not add to MLC licensing streams for z/OS, but the engines themselves are priced so compellingly that they really provide financial leverage in workload migration analysis.
As we would expect, System z continues to refine design elements to improve system performance for more kinds of workloads and applications. Every element continues to get refined from new chipsets and buffer strategies, to new connection technologies.
Look at the new range of workloads enabled with just with the faster chips on z10. Doubling the chip speed, while only a portion of system performance, does move the line to include compute intensive workloads that might not have been a good fit for movement or consolidation previously.
While it is easy to overlook, even new instructions are being created to help workloads. In the z10 these include instructions created to address floating point decimal functions, compiler code efficiencies (C++, PL1, and COBOL) and improved Java workload enhancements.
ISVs have been busy voting for System z as a platform. Consider 500 new ISVs last year joined the System z world. According to some of the numbers from our partner ISV community, there are 1300 new applications being built that will run on System z. This is in addition to the 1200 applications already enabled for Linux on System z, and the 4000 or so already available.
No new workloads?! We haven’t even talked about the other specialty engines, the expanding role of System z to deal with workloads related to energy management, administration and governance, serving enterprise data in-line, or new applications related to process and analytics that support new business models and could not have even been done a few years ago.
There was an article in the New York Times earlier this year that talked about the ‘mainframe’ and how it has evolved to be a whole new creature; unrecognizable from the thing we called ‘mainframe’. As it continues to evolve, the question we have to ask is not what workloads are there for it, but which ones don’t fit, and for how long?