high memory utilization..

[Shankar_6]

Hi,

I have a server running hpux 11.00.Memory tab in glance always shows 93%.Below is the output.

CPU Util SSNNAUU 37% 37% 37%
Disk Util FF| 32% 32% 32%
Mem Util SSUUB B 93% 93% 93%
Swap Util U UR R 73% 73% 73%

As we can see buffer usage,i could see dbc_max_pct is set to 50.

Below is the swapinfo usage

huedgpd1#swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 2512 0 2512 0% 0 - 1 /dev/vg00/lvol2
reserve - 720 -720
memory 4778 4625 153 97%
total 7290 5345 1945 73% - 0 -

Can we reduce the value of dbc_max_pct..? Will that have any impact..?why buffer cache is always in use..?

Points will be allotted.

- Shan

[Pete Randall]

Standard recommendation for dbc_max_pct is to tune it down to around 400MB (300 to 700 would be good).


Pete

[G V R Shankar]

Depends,

If you have Oracle or SAP, please reduce it 8.

kctune dbc_max_pct=8

Ravi.

[Steven E. Protter]

Shalom Shan,

Oracle does not benefit from buffer pool being large on this OS, which is out of support.

Drop it as low as you can get away with.

I set dbc_min to 5 and max to 7.

http://www.hpux.ws/?p=6
These tools will help you find the problem. I'm interested to see how they work on 11.00
If I'm not mistaken buffer utilization will be reported with this tool.

If you have a memory leak, a common oracle problem, this tool will help you find.

http://www.hpux.ws/?p=8


SEP

[yulianto piyut]

for memory above 8 GB, you should reduce the dbc_max_pct and dbc_min_pct. the amount of dbc_mac_pct for physical memory above 8 GB in the range under 1 GB. I had experience on it.
may be you can try for your server.

[Venkatesh BL]

buffer cache is used to hold parts of the file contents that has been accessed by your processes. As you have set the max_pct to a high value, there was no need to free up that space. If you reduce the max_pct, the excess buffer pages would be freed for other VM use.