Phil Payne writes: >Bollocks. >I've seen an IBM internal analysis of a Websphere Application Server implementation that was >37x cheaper on Intel than on zSeries. >That's 37 _TIMES_ - not 37%!
That's fascinating, Phil, but it's a hard fact that, today, the minimum order pricing for WebSphere Application Server for z/OS is less than the minimum for WebSphere Application Server on any other platform. Which was the only point I think I made, and it's simply irrefutable. Look at the IBM announcement letters and price it out if you don't believe me. Compare the price of WAS for z/OS's license and subscription/support at 3 MSUs versus 100 Value Units (minimum order for any other platform). And that's even leaving aside the fact that 100 VUs isn't a real world minimum order (unless you don't care at all about qualities of service, testing, etc., with the possible exception of Linux on z where you can still pull that off in 100 VUs) while 3 MSUs is a legitimate minimum. Actually, I have the price handy for 3 MSUs... Let me find it... The U.S. price is $2,300 one time charge for the license and then approximately $350 per year for subscription and support. Of course there's no charge for any zAAP capacity mated to those 3 MSUs, so with zAAP on average you end up with roughly 11 or 12 MSUs of total WAS capacity. Additional capacity beyond 3 MSUs has a sublinear price curve (or segmented sublinear polygon edge to be precise). The second segment starts at the 4th MSU, as a matter of fact. This same phenomenon is true for other products. WebSphere Message Broker for z/OS, WebSphere Portal for z/OS, and WebSphere Process Server for z/OS are some more examples. In principle a software developer or training provider that wanted to support/train across a large number of IBM software products could install everything in one 3 MSU LPAR and pay much less for a rather long list of products than they would if they did the same thing on a single X86 Intel/AMD processor. It's just the way it works: the mainframe has the lower entry software license price for much of the cross-platform IBM software catalog. Wasn't true some time ago, but it's true now. I've seen another cost study at another capacity (a large one) that shows WebSphere Application Server for z/OS is less expensive than other platforms. Knowing the source, construction, and verification of that study, I trust it. I suppose it's possible the 37X study was accurate for another set of circumstances, at another time, for another customer. For example, if I pretend zAAPs don't exist then that can dramatically affect Java economics. Or if your unrelated software vendors are going to "tax" you simply because you add capacity, even if it's a separate LPAR, that could be deadly. (Not IBM's practice. Thank goodness for competition.) Anything is possible. It's also true that WebSphere Application Server runs twice on mainframes: z/OS (or z/OS.e) and Linux. Each can have different economics depending on the circumstances. Finally, price is one factor. So are qualities of service. It depends what you're trying to accomplish. WebSphere Application Server for z/OS can deliver the best qualities of service. For many customers, for many applications, the price of NOT having these qualities of service dwarfs anything else. Sometimes, often, WAS z/OS is less pricey anyway, but sometimes, often, it doesn't even matter. - - - - - 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
