Hi ,
1. Regarding the device , what are the normal and the limit values I can expect from the controllers and the arrays ?

2. Is there more info regarding the different labels
Total IOs = max ?
Read Percentage = Better is higher ?
Cache Hit Percentage = Better is higher ?
Current KB/second = What are the device max ?
Current IO/second = = What are the device max ?

Thanks

Ok ... I've found a partial response for the Second point, but not the limit numbers

http://support.dell.com/support/edocs/systems/md3000i/en/CLI/HTML/maintsa.htm#wp1275257

I've got a few questions for you so we can help dig up the information.

1. What values are you asking for? Is it bandwidth out of the box and within the box? Can you be more specific?

2. How are you getting this data? Is it from the management console or OpenManage? Are you looking for the maximum values you expect to see or are you looking for typical values?

Can you tell us more about your setup as well?

Thanks!

Jeff

Good luck, I've been asking for the same info for 6+ months. Every time I talk with someone or see a doc, it says, 'well, it depends on what you are doing with it'. But no upper limit numbers on any of it...

I'm not sure that I fully understand what you mean by normal and limt values. In general the I/O and KB/s should follow normal rules for RAID disk sets. Can you elaborate just a bit on what you are looking for?

Thanks,
Todd

Sean,

We can help you but we need to know what values you are watching. Can you list out the values you want to know the maximum of?

In addition, we need to know the configuration. Believe it or not the maximums can be impacted by the exact configuration details. I hate to be a nudge, but we need to know loads of details.

Jeff

Thanks for your reply .

Yes , It depends .... but limits figures are good to know
I'm asking that because I've got a long and continuous output from (at least 10 MB of file to parse, av. 4 days of stats):

smcli -n md3000i -c "set session performanceMonitorInterval=10 performanceMonitorIterations=720;
save storageArray performanceStats file=\"c:\\md3000iPerfStats.csv\";"

But I don't know what can be the 'red line' to not cross ... If someone can tel me that xxx Kbs or IOs is the max level ... I"ll will take care about it.


Thanks

Here is the sample output of a couple of hours ...

http://www.flickr.com/photos/12269778@N05/?saved=1

OK, so you're asking for the IOPS and the Bandwidth. Is that correct?

Can you tell us your configuration? What does your MD3000i look like? What does you network look like? What about the clients mounting the LUNs?

Thanks!

Jeff

In Production we use Equallogic in Dev we use MD3000i

The current config is :

MD3000i (SATA) and MD1000 (SAS) and 5 ESXs.
Dedicated Gb Switches (Enterasys) and Gb Storage Interfaces (Intel).
Dedicated VMotion Gb switch (Enterasys)

The different LUNs are Raid5 and Raid0 (depending the requirements) ...
Snapshot are not activated
----
The Goal is to know if my performance issues are related to Disks ... or Controllers ...

I suspect throughput issues during ESX VM high and intensive load (Oracle data import). It slows down all the VM across the ESXs (no only one) .
For that I must know what is abnormal (blocking) element (Controller, Disks, Array)

Js

To get a better understanding of your config it would be nice to know the number of disks in the LUNs and the IO/s rates with those LUNs.

But - If you are seeing a slowdown that affects other LUNs, then that would point to either the controllers or the network or possibly the NIC on the ESX server.

Thanks,
Todd

Hi ... I've finished my script to analyze the data (with GnuPLOT and Linux).

You can find more info here : http://www.syldex.com/md3000i/

And a sample output ; http://www.syldex.com/md3000i/md3000i_Perf/result/index.html

I can conclude saturation on one controller.

Thanks for help.

Js