hudan
1 Copper

Latency and Less Troughput

Hi All,

My Customer told me to  check the disk below in vmax side

/dev/rdsk/c10t60000970000292601573533030453530d0s6

/dev/rdsk/c10t60000970000292601573533030453445d0s6

/dev/rdsk/c10t60000970000292601573533030453436d0s6

/dev/rdsk/c10t60000970000292601573533030453434d0s6

whether there is a latency, less throughput, the disks above
How to check it?
Thanks

Labels (1)
0 Kudos
6 Replies
Highlighted
dynamox
6 Thallium

Re: Latency and Less Troughput

you need to translate solaris disk to symmetrix device IDs, try:

symdev list -sid 123 -wwn | grep 60000970000292601573533030453530

and so on

hudan
1 Copper

Re: Latency and Less Troughput

i already found the device id , and how to check there is latency and less troughput?

Thanks

0 Kudos
dynamox
6 Thallium

Re: Latency and Less Troughput

you can look for service times in SPA or ControlCenter Performance Analyzer

Rasmmacias
2 Iron

Re: Latency and Less Troughput

Hello.

You can try:

symlv

 

Displays performance statistics and detailed mapping information about one or more logical volumes

that are defined in a logical volume group and performs control operations on logical volumes

 

To display statistics about all logical volumes of VXVM on SunOS every 30 seconds for one hour, enter:

 

symlv stats -i 30 -c 120 -type SUN_VXVM

 

The output key follows:

 

Volume Group Type : SunOS VxVM

H:M:S VgName LvName RW/s R/s W/s KbRW/s KbR/s KbW/s %Busy %Wait

A B C D E F G H I J K

A Time of day

B Volume group name

C Logical volume name

D Read and write requests per second

E Read requests per second

F Write requests per second

G KB read and written per second

H KB read per second

I KB written per second

J 100 * (logical volume active time / elapsed time)

K 100 * (non-empty wait queue time / elapsed time)

 

Quincy561
4 Tellurium

Re: Latency and Less Troughput

To me the best place to measure the throughput and latency would be on the host.  Why doesn't the customer look at iostat or sar?

For the Symm, most of the EMC folks have to rely on STP data from the service processor. Customers can use SPA or ECC.

0 Kudos
vforde
2 Iron

Re: Latency and Less Troughput

Additonally you could look at the symstat command for the device to check io metrics.

0 Kudos