I originally asked this question under the Oracle umbrella, but got no response, so I thought to ask you OS folks........
We have running our Oracle servers on Windows NT and 2000 servers. I am looking to try to identify, based on looking at the memory usage via task manager, how much memory is not in use (assume I am looking at it during peak times) so that I can safely add more memory to the Oracle resources. I understand that:
Commit Charge = total available memory (RAM + virtual)
Physical Memory = RAM
Kernel Memory = reserved RAM in use
My numbers show:
Commit Charge:
Total = 255M
Limit = 1876M
Peak = 264M
Physical Memory:
Total = 755M
Available = 426M
Sys Cache = 532M
Kernel Memory:
Total = 46M
Paged = 30M
Non-paged = 16M
1) how do the available and system cache number related to the total under physical memory. obviously the 2 don't add up to the total.
2) my available memory shown under commit charge shows only 255M available, some of which would be virtual memory. How can I find out what part of it is available physical memory, from which I can reallocate to Oracle resources?
3) is it normal that such a large portion of kernel memory is paged?
Note that I am the dba not the sysadmin so my understanding of this is at a very high level. Unfortunately the sysadmin is not much better with this, so here I am.
Any help would be appreciated. I have read a bit on the internet, but am still confused.
Thanks, Laura
Software/Hardware used:
ASKED:
June 4, 2004 8:19 AM
UPDATED:
June 4, 2004 4:06 PM
I’m agree: Performance Monitor is valuable tool in the case. Before tune up you should outline performance profile for application on particular box.
To figure out what counter means what in PerfMon, use its Help guide. It’s pretty good explanation.