> Some of the "squatty boxes" are blindingly fast at raw compute power, > blasting MF engines right out of the box. But how many business > applications are just raw computing?
Mathematically speaking, O(0) > the ones I've ever seen require significant amounts of I/O, and MF is > KING for I/O, because of concurrency. That's the really surprising thing for most folks. Those same blazingly fast boxes need lots of I/O to feed the beast. With lots of caching and multiple levels of caching, mechanical differences are disappearing. The open systems technology approach is obviously very different than the mainframe I/O architecture, but it gets the job done amazingly well and so you might not want to pick that fight either. Ron Hawkins might chime in with some pithy observations on relative I/O performance. > And running multiple concurrent tasks on a "squatty box" rapidly bogs > down, no matter how efficiently the software is written. That's not really true anymore. It is true that average utilization of typical servers is relatively low and that their processor designs are typically less friendly towards high context-switch rates than z. Their architectures are not necessarily "efficient" in the classical sense of keeping the processor as busy as possible, but they are unarguably effective and cost effective too. > So in terms of USEFUL work, MF is still The Big > Cahuna. Some day, that might change, but I don't recommend holding your > breath waiting for it. Fortunately for me I don't need to hold my breath on this one. > The bottom line is, and should be, what satisfies my business needs in > the most timely and cost-effective fashion. Agreed. CC ---------------------------------------------------------------------- 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
