Network and Storage Protocols

sysstat and statit

bipul34_gogoi
17,073 Views

Hi all ,

I am trying to find out some relation ( formula !!)  . First , for sysstat ...

CPU   NFS  CIFS  HTTP   Total    Net kB/s   Disk kB/s     Tape kB/s Cache Cache  CP   CP Disk    FCP iSCSI   FCP  kB/s
                                                    in   out   read  write  read write   age   hit time  ty util                 in   out
32%   430     0     0          430  1923   128    348    972     0     0    32  100% 100%  Bv  75%      0     0     0     0
20%   386     0     0          386  1725   116    256    612     0     0    32  100% 100%  Bv  73%      0     0     0     0

Whats the relation between TOTAL IOPS and Disk KB/s (Read+write) , Above output is from a load testing with 4KB IO.. (1923+128/430= 4.76)

Secondly , How those relations can be related to invidual disk IOPS

Statit o/p

disk             ut% xfers  ureads--chain-usecs writes--chain-usecs cpreads-chain-usecs greads--chain-usecs gwrites-chain-usecs
/aggr4/plex0/rg0:
v5.26             15   6.37    0.09   1.00 365417   6.19   2.03 13595   0.09   4.35  5600   0.00   ....     .   0.00   ....     .
v5.27             16   6.57    0.09   1.00 450435   6.39   2.03 13724   0.09   3.68  6914   0.00   ....     .   0.00   ....     .
v5.28             19   8.10    1.21   1.00 52718   4.87   1.69 20546   2.03   2.65  8867   0.00   ....     .   0.00   ....     .
v5.29             14   5.76    1.18   1.00 49934   2.28   2.50 10604   2.31   2.21 12602   0.00   ....     .   0.00   ....     .
v5.32             14   5.76    1.18   1.00 51887   2.28   2.50 10646   2.31   2.22 12357   0.00   ....     .   0.00   ....     .

Note : v5.26 and v5.27 are parity disk .

From documeantation i can see that Xfers is "total number of disk ops" .  Whats the relation (formula) between Xfers from statit o/p and Total IOPS from sysstat command ..

Thanks in advanced

4 REPLIES 4

BrendonHiggins
17,071 Views

The statit is the individual disks so you need to add up (ignore parity disks) all the results from each of the raid groups to see what is going on.  The sysstat is a total for all the IO and if you have just one aggregate life is simple.  If you have multiple I think it shows the information for the busiest aggregate and sum total in other areas.  Do not forget to take note of chaining as sequential data can have multiple reads/writes during a single IO event.  The total is number of read or writes, multiplied by the chain, multiplied by 4k (block size).

How this helps but it has been a long day and I think I should leave others better able to explain.

Bren

bipul34_gogoi
17,070 Views

Here we have only 1 aggr ..

Do you mean that ..

Total IOPS from sysstat output = sum of XFR from statit o/p  of the data disks  ??????

Regarding the chain .. whats exactly the value you are refering from the statit o/p .. I can see  many chain values . like (ureads--chain-usecs, writes--chain-usecs .......)

Thanks for responding ....

BrendonHiggins
17,071 Views

If you have a look at the slides it tells you what you need to know.  http://communities.netapp.com/people/BrendonHiggins/blog/2009/10/21/sept-nug-meeting

This is also a very good link

http://communities.netapp.com/docs/DOC-1051#cf

Hope it helps

Bren

Cloud
14,992 Views

can we get this out via report using Operation manager?

Public