Stephen,
> Hi All,
> Another item for tomorrow if you have them available is
> reliability numbers.
What is the basis for the reliability number? Mean time
between reboots (which looks like what you are doing
below)? Percentage of hours available per day?
If one includes scheduled outages, I would say one
sees (about) 3-5 major patch cycles per year for
security and system updates (as always, your mileage
will vary). Which, if you look at it that way, means
somewhere around 90 days between reboots (which would
seem to suggest a 1% reliability number based on the
following estimates).
> I remember for the VA Linux machines (sorry to open
> old wounds) that they typically needed rebooted once
> every 30 days, which put a 3% (roughly) needing
> rebooting per day. What is a similar number for
> other linux machines at SLAC? At CERN they claim
> the normal number is 1% which seemed kind of high
> to me. They also said that Fermilab had a similar
> number.
Gary
|