EWR@VAX01.AMS.COM (Betsy Ramsey) (08/02/88)
I just posted a separate message asking about tools for capacity
planning, VPA in particular. This is a more general question.
For those of you running larger VAXs (8600, 8700), what combination(s)
of CPU and I/O usage do you find produces "unacceptable" response
time? This is really several separate questions:
o Do you measure response time? If so, how? (Do you use one of the
"load average" drivers? Do you run your own home-grown tool?)
o How do you perform capacity planning? That is, can you accurately
predict (preferably six months or more in advance) when you will
need to purchase additional hardware (disk drive, channels;
memory; CPU)? If so, what tools and methodogy do you use? I'm
particularly interested in the methodology.
o How do you measure your workload? That is, what tool(s) do you
use to know, on average, which users (or accounts) running which
images are responsible for using your system resources?
If you are successfully answering these questions (particularly the
last two), I'd appreciate hearing from you. If you'd rather not write
it all down, I'd be happy to give you a call.
Betsy Ramsey EWR@VAX01.AMS.COM
American Mathematical Society Phone: 401-272-9500
Providence, RI
-------