wren2000
asked on
Memory State of Health
Sun Solaris E250 running SunOS 5.6
512 MB Physical Memory
Approx. 1.5 GB Swap Space
I need to determine the "state of health" of memory on this server. I have run many monitors which includes multiple variations on the vmstat and sar commands. Unfortunately, I'm not a UNIX admin, and I am trying to determine what the results mean. I have included the results below.
Here are my questions:
It looks to me like swapping is not a problem on this server. Paging does occur, but isn't some paging normal? What is a normal range? Am I out of the normal range, and must I increase physical memory on this server? Are 188.06 address translation faults per second excessive, and what can reduce that number?
Swapping:
vmstat 5 10 and sar -w 5 10 indicate 0 swapping
Paging:
vmstat 5 10 indicates
==> 0-12kb pi
==> 1-60kb po
vmstat -S 5 10 indicates
==>1-38kb pi,
==>4-22kb po
sar -g 5 10 indicates an avg of
==> 2.84 pgout/s,
==> 2.86 ppgout/s,
==> 4.14 pgscan/s
sar -p 5 10 indicates an avg of
==> 1.64 pgin/s
==> 1.68 ppgin/s
==> 188.06 vflt/s
sar -r 5 10 indicates an avg of
==> 1057 freemem
==> 3,236,369 freeswap
Thanks.
512 MB Physical Memory
Approx. 1.5 GB Swap Space
I need to determine the "state of health" of memory on this server. I have run many monitors which includes multiple variations on the vmstat and sar commands. Unfortunately, I'm not a UNIX admin, and I am trying to determine what the results mean. I have included the results below.
Here are my questions:
It looks to me like swapping is not a problem on this server. Paging does occur, but isn't some paging normal? What is a normal range? Am I out of the normal range, and must I increase physical memory on this server? Are 188.06 address translation faults per second excessive, and what can reduce that number?
Swapping:
vmstat 5 10 and sar -w 5 10 indicate 0 swapping
Paging:
vmstat 5 10 indicates
==> 0-12kb pi
==> 1-60kb po
vmstat -S 5 10 indicates
==>1-38kb pi,
==>4-22kb po
sar -g 5 10 indicates an avg of
==> 2.84 pgout/s,
==> 2.86 ppgout/s,
==> 4.14 pgscan/s
sar -p 5 10 indicates an avg of
==> 1.64 pgin/s
==> 1.68 ppgin/s
==> 188.06 vflt/s
sar -r 5 10 indicates an avg of
==> 1057 freemem
==> 3,236,369 freeswap
Thanks.
ASKER
I appreciate your response...
1) This server is an Oracle database backend for a web site. I wanted to determine the state of memory on the box because I have consultants telling me they will be adding 61 application connections to the system compared to the 14 they have now. I really need to know whether or not the system in its current state can handle that many additional processes, and I wanted to start with how well it is handling the current load.
2) I tried to run vmstat -p, but that switch isn't recognized on my system. Only -cisS are recognized.
Thanks!
1) This server is an Oracle database backend for a web site. I wanted to determine the state of memory on the box because I have consultants telling me they will be adding 61 application connections to the system compared to the 14 they have now. I really need to know whether or not the system in its current state can handle that many additional processes, and I wanted to start with how well it is handling the current load.
2) I tried to run vmstat -p, but that switch isn't recognized on my system. Only -cisS are recognized.
Thanks!
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
No comment has been added lately, so it's time to clean up this Topic Area.
I will leave a recommendation for this question in the Cleanup topic area as follows:
- Answered by chris_calabrese
Please leave any comments here within the next 7 days
PLEASE DO NOT ACCEPT THIS COMMENT AS AN ANSWER !
tfewster
Cleanup Volunteer
I will leave a recommendation for this question in the Cleanup topic area as follows:
- Answered by chris_calabrese
Please leave any comments here within the next 7 days
PLEASE DO NOT ACCEPT THIS COMMENT AS AN ANSWER !
tfewster
Cleanup Volunteer
per recommendation
SpideyMod
Community Support Moderator @Experts Exchange
SpideyMod
Community Support Moderator @Experts Exchange
Next, more memory will almost _always_ make a Solaris machine faster. That's because Solaris will use any available memory as a disk cache.
As for swapping vs. paging, the default Solaris behavior is to only swap when there is a severe memory shortage. So zero swapping indicates that you don't have one.
Finally, to see if your programs are actually getting squeezed for memory, you want to run vmstat -p so you can see a breakdown of the different paging data. In particular, you want to see that apo and epo are low, as the others could be high on a busy system, even when there's no memory shortage.
If apo and epo are much higher when the system is having performance issues, then you need more memory.
Curiously, however, the sol-7 boxes I have document the -p switch but don't implement it...