>Now for some heresy; one ICF should suffice. The mean time to failure of an >engine in this scenario is probably measured in units of centuries.
Yes! But! I worked for a company, many moons ago, that relied on that philosphy! Now, the horror story begins: We had had 3 failures on a 9672-R83 within 6 months. They were supposed to be reliable. IBM was in major damage control after the third one. They gave us a major presentation showing that there were just over 1200 (something like +17) Rx3's installed world-wide. Only two had had more than one failure, and only one had more than 2. During that presentation, we got paged. Suddenly, only one box had more than 3 failures. Ours. IBM's damage control? They did everything they could to get it physically replaced, rather than just upgrade in place. The reason: To help the customer? No! To improve their stats! They couldn't move fast enough to give us a deal on an R35, and take the other box out. They probably ground it into dust! My point? Schmidt happens. Accepting the so-called 'high availability' as a planning option, is risky. A year ago, I had two SYSPLEX timers fail within minutes. Not because of hardware. Because they were each upgraded to a new level of microcode, one after the the other. Then, within minutes, after they were back online, BINGO! BANGO! BONGO! The whole 'PLEX was down! Also, remember that an availability of 99.5% means an unavailability of 0.5%. Factor that into the number of devices installed. Something will go down every day. In other words, it may be hype/marketting, but it's also an insurance policy. Pay for the best you can afford, which also includes risk management. Which costs more? The outage? The cost to prevent that outage? PS: I always have gone with at least 2 CF's in production. When in doubt. PANIC!! ---------------------------------------------------------------------- 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
