Frank Sabo writes: >I currently have an IBM7060 H75 processor running OS 390 >2.10. We are currently going into our sixth year of a two >year phase out of the mainframe.
As you mentioned, that system has roughly 209 MIPS. It's a G5 class processor. Physically it is pretty much equivalent to a 9672-R26 model, if you're curious. (I don't think this model had internal storage, so that's not relevant here as far as I know.) Yes, your employer can do much better, especially if they can adopt a very rational, business-centric view. In other words, it may be a good time to question dated assumptions and preconceived notions about what you can do with a 2007 mainframe. Everything has changed in 6 years. Some ideas follow below. >Since everything is non supported management is looking >into upgrading the processor to a 2096 R07 (F02) box and >operating system to the latest ZOS version, since it now >looks like it would be another two to three >years before can be converted off. If I can come up with a list of >jufitications for upgrading the processor and operating system. The >system then might be taken off the endangered species list. The 2096-F02 you are considering is rated at approximately 170 MIPS. Now, I don't know what software licensing you have or what products you have, but today you would be paying full capacity software licensing at 209 MIPS and the equivalent of about 37 MSUs, probably. At full capacity a 2096-F02 is 24 MSUs, and if you start sending in your SCRT reports and softcapping your LPARs you could end up with much less than that. [Oddly enough, if management claims they're moving work off the mainframe in a couple or three years, that claim reinforces the benefits of VWLC licensing: the software charges decline (somewhat) along with the workload shift. (Of course, the target platform is probably more expensive to buy, own, and operate. If they're fixated on software license prices for some reason rather than how much work actually gets done, then VWLC is a good thing.)] If you can get the mainframe-to-mainframe migration done within 90 days, you can start paying the new software rate from the first day forward. You'll also get one year of warranty (i.e. no charge for hardware maintenance), so you avoid 12 months of that. Once you do start paying hardware maintenance again it might be a lower rate. There are various IBM promotions for this sort of migration and for DS6000 storage. Ask. You'll avoid a hugely expensive migration to another platform. (Did anybody bother to fully cost that? :-)) >1. SOX hardware and software compliance. Yes indeed. Don't forget crypto support: it's quite excellent on the z9, so you can get stronger network security if you do have to make a network hop for something. You also have potential benefits with zIIP (certain DB2 workloads and increasingly other workloads) and zAAP (Java and increasingly other workloads). >2. All software will be vendor supported. And this support actually is support. How refreshing. :-) >3. Installation of ZVM which would allow us to run LINUX >on the box. Yes, absolutely. Ask about any promotions if you do this on the same purchase order. >4. All new applications to be considered to on the platform. And this is the key. What "new workload" would fit? If you can find even a handful of Oracle licenses to put on a Linux processor, for example, you're making more money and improving service. One thing the bosses really need to understand is that the 171st MIPS is much, much, much less expensive than the first. This is a scale platform. So how much does it cost to, for example, add a little bit of WebSphere Application Server in a small zNALC z/OS LPAR to a mainframe? Answer: it's less to acquire than any other system, and less to operate. Another factor that seems to be driving many businesses to expand their mainframes (vertically): data center space. If your data center is full (physically, electrically, and/or cooling-wise), or getting near full, building a new data center is quite expensive. I'm not sure how your employer is doing, but a walk through the data center(s) should be instructive. Re: Making the migration itself, there are various ways to do it. One way is to get your current system up to z/OS 1.4, then bring everything over using z/OS 1.4 Bimodal on the z9 (to get the "technology dividend") as fast as you can, then flip the switch to 64-bit as fast as you can (to get VWLC), then go to z/OS 1.7, then normal currency cadence to z/OS 1.9 (when available) and beyond. The faster you can get onto the z9 the better the immediate financial picture, I suspect. Concurrent with the 1.7 upgrade you can get your middleware (DB2, MQ, IMS, CICS, etc.) all current. (DB2 9, CICS 3.2, etc. all run on 1.7.) This is one recipe, but you'll want to plot this out to figure out your own best approach. Hope that gets you started. Sounds like a really good idea, and it should be a strong business case, especially if you can turn the corner on data center, application, and IT staff rationalization. Also check out Hannaford Brothers and the stories about their System z9 upgrade, another grocery company. (You can find the stories via Google.) - - - - - Timothy Sipples IBM Consulting Enterprise Software Architect Specializing in Software Architectures Related to System z Based in Tokyo, Serving IBM Japan and IBM Asia-Pacific E-Mail: [EMAIL PROTECTED] ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
