On Mon, 2007-01-22 at 20:05 +0100, R.S. wrote: > David Andrews wrote: > [...] > > Some years ago we had a mixture of servers running NetWare [...] > > raised a NMI [...] on the average of once a month
> Once a month? What hardware did you use??? Those were (IIRC) HP Netserver LEs and LFs. I did say it was awhile ago! > There were failures, both hw and sw, but > the frequency was *significantly* lower. Lucky you. Certainly I am comparing apples and oranges; the parity machines were different from the non-parity machines. Bit errors on one system might-or-might-not be as frequent as bit errors on another. But you work with the data you have, and crude extrapolation told me that once a day we had uncorrected bit errors on one of those 300 machines. That meant that (if we were lucky) somebody's machine locked up, causing us to spend a half hour of a tech's time examining the carcass looking in vain for a software issue that wasn't there, and the user had to recreate whatever document was in-flight at the time. I say this is the lucky case, because the costs of recovery are well-defined. But in the unlucky case the bit error is undetected and makes a mess in someone's spreadsheet, or corrupts a buffer, or... what? You'll never find out what that data error cost your company. Take my dubious numbers as you will, but my point was that bit errors do happen. Insurance is a good thing. -- David Andrews A. Duda and Sons, Inc. [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