roy@phri.UUCP (Roy Smith) (06/27/89)
I've been keeping an eye on our 4-Mbyte vax-11/750, trying to
figure out if it would really help things if we added more memory. The
idea is that if you do a lot of paging, you need more memory, but how do
you know what "a lot" of paging is? For example, right now I've got a Sun
perfmeter running looking at paging on the 750 (we run rstatd on the vax)
and it just flipped over to the 64 scale. Is 64 (1k) pages/second "a lot"?
Over the past few minutes, it's been at or near 0 for perhaps half the
time, with a few peaks up in the 30-50 range. Is that "a lot"?
Other than performing the experiment of buying more memory and
seeing what happens, how does one know if you need more or not? I'm
inclined to try the negative control experiment by taking one or two of the
4 1-Mbyte cards in there now off line and seeing what happens, but I
suspect people would lynch me.
--
Roy Smith, Public Health Research Institute
455 First Avenue, New York, NY 10016
{allegra,philabs,cmcl2,rutgers,hombre}!phri!roy -or- roy@alanine.phri.nyu.edu
"The connector is the network"grr@cbmvax.UUCP (George Robbins) (06/28/89)
In article <3818@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > > I've been keeping an eye on our 4-Mbyte vax-11/750, trying to > figure out if it would really help things if we added more memory... > ..Is 64 (1k) pages/second "a lot"? > > Over the past few minutes, it's been at or near 0 for perhaps half the > time, with a few peaks up in the 30-50 range. Is that "a lot"? It depends on your work load. If you have some mega-tasks that simply won't fit in memory, it's no big deal. On the other hand, if it's some interactive task that has to be paged back in before the guy gets a response to hitting return, or bopping some editing key then it might seem pretty horrible. I'm sure you'll get some horror stories from some of the educational timesharing folks, but I'll suggest a simple alternative... Used 750 memory boards (DEC) go for maybe $600/M-byte, you should be able to get 4 for $2K with only a little haggling. Do it, a 750 is painful enough without adding wasted activity. I'd think 8 M-byte is about right, unless you have a lot of tasks like emacs, lisp or other virtual memory eaters. Seems that 16-MB in a ~32 user 785 doesn't normally page at all, and you're about 1/3 to 1/2 of that. You might also want to look at the prices for used Microvax II's or the like. Unless you have some profound affection for that 750, you should be able to do much better without spending much. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@uunet.uu.net Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
dik@cwi.nl (Dik T. Winter) (06/28/89)
In article <3818@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > > I've been keeping an eye on our 4-Mbyte vax-11/750, trying to > figure out if it would really help things if we added more memory... > ..Is 64 (1k) pages/second "a lot"? > > Over the past few minutes, it's been at or near 0 for perhaps half the > time, with a few peaks up in the 30-50 range. Is that "a lot"? Well, this weekend I had a 12 hour session with a CDC Cyber 995 that gave a total of 11,000,000+ page faults, or about 260/sec. And the machine was not very loaded. ;-) Oh, and page size is selected at deadstart and is 2K, 4K, 8K or 16K words (8 byte words) in size. Must have been some disk traffic. -- dik t. winter, cwi, amsterdam, nederland INTERNET : dik@cwi.nl BITNET/EARN: dik@mcvax