On Thu, 12 Dec 2013 09:41:19 -0500, Scott Ford wrote: >I wasn't aware of Cobol SRBs violated IBM licensing. I don't mean to be >ignorant, by why ?
ITYM "running Cobol on a zIIP", not "Cobol SRBs". You have to understand a little about software pricing. When IBM started to charge for MVS, it seemed to make sense to charge more for a bigger processor. The general idea was that the software charges would be almost twice as much on a processor that is twice as fast. Actually, the idea was that a small shop shouldn't have to pay as much as a big one. After a while, there were no more of the smallest model group processors available. IIRC, the pricing was never linear, but not terribly far from it. In those days a big computer would run at about 4 MIPS and you could license MVS on a smaller processor for about $2000 per month. With the latest mainframes approaching 80,000 MIPS, the price of the operating system is many times higher than it was 30 years ago. It is not uncommon to pay hundreds of thousands of dollars per month for MVS. The result has been that other operating systems are considered to be preferable in any case where the other operating system seems to be adequate for the purpose. Consequently, new workloads are rarely deployed on MVS. This has been true for a long time despite the fact that many of these new workloads could run just fine on MVS, and benefit from the strengths of MVS and the mainframe. And even arguably be less expensive to run on MVS. While the use of computers has increased at an amazing rate over the last few decades, the installed base of MVS systems has not kept pace, and has instead declined. Actual numbers are hard to come by, but we all know sites that were once MVS sites that now use other systems instead. Does anyone know of any new MVS sites? As processors continued to get faster, the pricing structures were revised several times. Parallel Sysplex pricing was the first clue that I saw that IBM recognized that software pricing was an issue. In fact, the talk about the mainframe being expensive is really about software pricing. With z/Architecture machines, IBM started to offer a "technology dividend" in the form of a relatively modest discount for the software charges on the newer processors. This was done by adjusting the MSU rating, so that a new computer that would run a given workload using the same amount of resources as the previous generation would have an MSU rating that was maybe 10% less than the equivalent processor of the previous generation. Since software charges are now based upon the MSU capacity, this reduced the charge for running a given workload on the newer processor. However, the newer generation of processor would top out at maybe 50% more capacity than the previous generation, so the potential cost of running z/OS continued to rise. One consequence of the software pricing problem is that for most of the models of z/Architecture systems, the processors have been deliberately slowed down, allowing customers to acquire mainframes with a capacity that is closer to their actual needs. As a result, often there is no spare capacity for new workloads.. zIIP and zAAP processors were introduced to allow customers to exploit new system facilities that require additional processing power without having to incur additional costs for the software. This was done by providing zIIP and zAAP processors that could run only certain kinds of work, and that would not count when calculating the compute power of the complex. Therefore, adding zIIP and/or zAAP processors doesn't increase the software charges for the complex. zIIP and zAAP processors are a short term solution for the problem of increasing software charges. The result of the increasing software charges is that customers minimize the use of the mainframe, and have been doing so for decades. One way of doing that is to move workload from MVS to other platforms. The other way is to deploy new workloads on other systems, without even considering the mainframe as a possible place to run them. In other words, as the size of the data center grows to accommodate more applications, the non-MVS hardware grows at a considerably higher rate than the MVS related hardware. Even in shops where no applications have moved off of MVS, the percentage of the workload that could run on MVS that does in fact run on MVS is decreasing. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
