>The prices for CICS and DB2 grow with version. No, not necessarily. Offload engines are one reason. DB2 V8 will support zIIPs, and DB2 V7 won't. V8 also has multi-row fetch/insert and MQTs. Does V8 have a higher price?
CICS 3.1 has better performance for thread-safe code and for SOAP (Web services) than prior releases. And any CICS Java can now offload to zAAPs. Does 3.1 have a higher price? To use an analogy, if a box of double strength laundry detergent costs 50% more, does the original formula really cost less? Of course your mileage may vary, but there are real, actual reasons why core product prices are declining for many shops (or growing slower than business growth). And that's leaving aside inflation adjustment. Software prices should, by all rights, be increasing at approximately the rate of wage inflation. While that's true of mainframe software (if you ignore all the effects I just mentioned), unfortunately distributed software has been increasing at a faster rate. >Last but not least: Majority of the prices are MSU dependant. Average >MSU per customer grow significantly. Ergo: even if prices remain >unchanged, the fees grow up. MSUs grow only if the business actually grows. I think most people would agree that, if your business gets larger, it's fair to pay slightly more for software. Otherwise Wal-Mart would pay the same dollar amount for software as Karen's Chocolate Shop, with "two locations to serve you" in greater Allentown. :-) There are a lot of reasons why charging MSUs can decline. For example, get a new mainframe model with the same computing performance and your MSUs will decline by 10%. Jump two models forward and the figure is -19% (i.e. the effect of two 10% reductions). Get an offload engine (crypto, IFL, zAAP, zIIP) and you could see a dramatic effect, depending on the workload. Move workload onto the mainframe to gain architectural proximity and you'll see a reduction in the MSUs consumed by the subsystem you're connecting to. (Basically DB2 works less hard with Hipersocket attachment and even less hard if the work originates in the same LPAR. The net effect depends on the workload.) Many shops have also discovered that WLC charging is beneficial. (If it isn't it's OK to stay on another formula.) There's been a tendency lately for increased granularity (smaller capacity steps) in the newer mainframe models, lately the z890. That helps manage software costs to more closely match exact business requirements. On-off capacity on-demand is relatively new, and if you only need a little bit of extra boost for a little bit of time (such as annual reporting), that's a great option for cost containment. Softcapping is also quite handy. IBM has injected an enormous amount of competition into the tools and utilities business in recent years, and that's had quite an impact. z/OS.e is a relatively new, lower cost offering for z800 and z890. z/VM prices have declined across the board as its popularity has surged (with Linux). All of the above is not to say that mainframe software isn't (still) "expensive." It is. But has anyone checked the prices of distributed software lately? Ouch! There's a reason why Microsoft collects far more software revenue than IBM, and it has nothing to do with mainframe software. Nor does it have much to do with high volumes of business transaction processing either -- there's an awful lot of money spent for relatively few transactions. - - - - - Timothy F. Sipples Consulting Enterprise Software Architect, z9/zSeries IBM Japan, Ltd. 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
