when i execute iostat 10 10 command i see the following this, i just want to know what are all these, i mean dasda-dasdq and dm-0 to dm-12.
we are using datastage v 8.1.1 and db2
Linux 2.6.16.54-0.2.5-default (vmzlsliis07) 10/20/2010
avg-cpu: %user %nice %system %iowait %steal %idle
2.00 0.00 0.48 1.50 0.51 95.52
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
dasda 1.52 16.80 37.09 8164704 18028984
dasdb 1.52 2.31 26.55 1121408 12903048
dasdc 0.14 5.79 1.02 2816000 494712
dasdd 0.02 0.21 0.39 100744 187712
dasde 0.00 0.00 0.00 432 0
dasdf 0.00 0.00 0.00 432 0
dasdg 0.00 0.00 0.00 432 0
dasdh 0.00 0.00 0.00 432 0
dasdi 0.00 0.00 0.00 432 0
dasdj 3.78 292.75 74.35 142285368 36137736
dasdk 1.79 303.77 19.65 147638808 9550816
dasdl 0.22 145.96 8.66 70942560 4210368
dasdm 0.65 59.23 13.69 28789896 6652816
dasdn 0.12 43.65 3.76 21216688 1828296
dasdo 0.00 0.01 0.00 5536 0
dasdp 0.00 0.01 0.00 5536 0
dasdq 0.00 0.01 0.00 5680 0
dm-0 8.69 278.93 65.00 135569456 31590048
dm-1 1.90 296.22 12.15 143969312 5903784
dm-2 1.34 167.23 8.67 81279136 4211704
dm-3 0.89 20.46 4.59 9943624 2229816
dm-4 0.01 4.21 0.00 2048112 1592
dm-5 0.99 13.79 7.67 6700632 3727632
dm-6 0.98 13.26 7.66 6444176 3724808
dm-7 2.26 51.18 11.27 24875112 5475328
dm-8 0.21 0.02 1.69 9888 820056
dm-9 0.18 0.01 1.43 6336 695264
dm-10 0.07 6.37 0.41 3097632 201240
dm-11 1.54 0.02 12.29 8720 5972600
dm-12 1.91 1.71 14.86 828800 7223920
Any idea on these, will be really helpful.
iostat 10 10 command output
Moderators: chulett, rschirm, roy
</a>
iostat measures disk input output. The disks represent filesystems mounted on your box. You have a lot of blocks being written. The command works off of snapshots or a moment in time. You told the command to take 10 snapshots , one every 10 seconds.
vmstat is a similar command. You can look at whether or not a system is using a lot of RAM. The older version of these commands is the sar command. In UNIX there are a couple things to look for. Are you paging? This means you have run out of RAM and program pages are being swapped out to disk. Disk drives are slower than RAM so this slows down your processing.
Back to the iostat, what is most important is are you waiting on any filesystem. If you have iowait then the system has to wait for one drive to write out its disk buffers. This is a bottleneck. If you spread your files across more filesystems then you might not be waiting on the one filesystem. If the filesystem has /bin and /usr/bin then you are waiting for the UNIX commands to be loaded into memory.
vmstat is a similar command. You can look at whether or not a system is using a lot of RAM. The older version of these commands is the sar command. In UNIX there are a couple things to look for. Are you paging? This means you have run out of RAM and program pages are being swapped out to disk. Disk drives are slower than RAM so this slows down your processing.
Back to the iostat, what is most important is are you waiting on any filesystem. If you have iowait then the system has to wait for one drive to write out its disk buffers. This is a bottleneck. If you spread your files across more filesystems then you might not be waiting on the one filesystem. If the filesystem has /bin and /usr/bin then you are waiting for the UNIX commands to be loaded into memory.
Mamu Kim
If /tmp is also on this filesystem then move it. You could also separate /bin from /usr or some other directories to split this load up into different filesystems.
A lot of UNIX admins create all filesystems from one virtual filesystem. Meaning all physical disk drives are lumped together into one big virtual drive and filesystems are created from the virtual drive or volume. Sometimes drives are called volumes. All means the same thing. This means splitting these filesystems may not improve performance because they share the underlying disk drives. Oldtimers always wanted to control which physical drive gets hit so they could improve performance when needed. Some admins just want the easiest way to admin. Blame the hardware. So when you need more performance you need a faster computer. Just my opinion. See it too often. Almost all LAN drives are lumped into one big virtual drive. So they can never improve the performance.
A lot of UNIX admins create all filesystems from one virtual filesystem. Meaning all physical disk drives are lumped together into one big virtual drive and filesystems are created from the virtual drive or volume. Sometimes drives are called volumes. All means the same thing. This means splitting these filesystems may not improve performance because they share the underlying disk drives. Oldtimers always wanted to control which physical drive gets hit so they could improve performance when needed. Some admins just want the easiest way to admin. Blame the hardware. So when you need more performance you need a faster computer. Just my opinion. See it too often. Almost all LAN drives are lumped into one big virtual drive. So they can never improve the performance.
Mamu Kim
Thanks a lot
Thank you all for your Replies. ya during morning times I observed that the iowait of our server is almost reaching 98% usage and during that time even logging into a datastage client is taking a very long time. and some times every one who is logged into this server are getting kicked out of the clients and the putty too. so I want to know what exactly is using the io that much. when i executed the iostat 10 10 i saw these variables but dont know what are those, so thats why i posted here. I requested our unix team and they figuered it out that online backup is scheduled on the server every day morning 10AM, so it is using all the io. so we asked them to change the timing and add more CPU's to our system that resolved the issue. Thanks a lot folks