performance checking by using time

[ricky2]

Hi,

Iam working on 2 hp servers of model 9000/800/L3000-8x which are having the same file system and both these servers are being replicated.
Iam having a performance problem on one of them.
when I try to check a extraction of a same zip file on both server1 and server 2 using the time command.
server1>time unzip xyp.zip
real 1:26.1
user 1.5
sys 2.3

server2>time unzip xyp.zip
real 20.3
user 1.7
sys 1.7

you can observe there is a huge diffrence in the time between server1 and server2.
Even the disk utilisation for server1 is much lesser than server2.
could anyone help me out on this problem?
thanks in advance

Rgds,
Ricky

[Muthukumar_5]

check system resource usage as,

# top
# vmstat
# iostat

It will give you the way to tune. Hope the system configuration on swap, memory, processor are same.


--
Muthu

[Arunvijai_4]

Hi Ricky,

You need to check various stuffs in the system,

# top
# vmstat 5 5
# iostat

will help you. Also, SEP's system monitor script provide you with many details.

http://www.hpux.ws/buildmail.hpux.text


-Arun

[Arturo Galbiati]

Hi Ricky,
did you are using the same unzip program version and xyp.zip file?
HTH,
Art

[James R. Ferguson]

Hi Ricky:

You have posted one measurement and at that for a process that runs relatively quickly. You have little idea of how *other* processing, including I/O rates, queue depths and transfer times are impacting *your* process. A better analysis would begin with 'sar' and 'glance' performanace measurements. Overall, over time, there may be little real difference.

Regards!

...JRF...

[ricky2]

Hello,

I have allready analysed using the sar,iostat and top and what I analysed is that the server1 which has considerable bad performance has good results of this and
below are few results

Average SAR

server1:
Average 6 8 1 85

server2:
Average 9 10 1 80

also iostat and vmstat could not help me out in identifying the difference.
I have also tested other processes on the two machines and found both are accomplishing the job alomost at the same time.
How can I check the queue depth?
Iam using the same unzip version and testing on the same file on both the machines.

Rgds,
Ricky

[Ivan Ferreira]

As you can see the user and sys time are very close, this means that the system is spending more time on other kernel functions.

You should check the load average for both servers.

[Steve Steel]

Hi

If you are really stuck get a trial copy of this http://www.sarcheck.com/index.htm


Seems to me they are not identical and one is doing a lot more disk io so slowing the rest


what does
w -l

Give on the 2 machines

Steve Steel

[ricky2]

Hi Steve,

Thanks for the link

The w -l out of the two machines is as follows

server1:
4:30pm up 61 days, 18:17, 3 users, load average: 1.58, 1.56, 1.56

User tty login@ idle JCPU PCPU what
ren pts/ta 09:35am..................

server2:
4:35pm up 61 days, 18:31, 2 users, load average: 0.18, 0.23, 0.25

rgds,
ricky

[Shane Russell]

Ricky,

Do you have GlancePlus Pak installed?

Are you using PerfView or Performance Manager?

Shane