Standard Benchmarks for File Servers
 

What are considered the standard programs for benchmarking performance of a
file server these days? Ideally I would like to have some measurements not
just on throughput, but on latency on trivial I/O operations like expanding
a share folder. Of course I would want to see some graphs at end of test
to show a relationship between the performance metrics and numbers of
simulated users.

--
Will

Standard Benchmarks for File Servers
 

"Will" wrote in message
...
What are considered the standard programs for benchmarking performance of
a file server these days? Ideally I would like to have some measurements
not just on throughput, but on latency on trivial I/O operations like
expanding a share folder. Of course I would want to see some graphs at
end of test to show a relationship between the performance metrics and
numbers of simulated users.


For disk performance, try I/O Meter from www.iometer.org

It's the tool HP will get you to use if you claim any suspect disk
performance. I haven't used it myself, but it is scriptable.

Standard Benchmarks for File Servers
 

IOMeter is targeted at testing and benchmarking locally attached drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.

--
Will

"Jez T" wrote in message
...

"Will" wrote in message
...
What are considered the standard programs for benchmarking performance of
a file server these days? Ideally I would like to have some
measurements not just on throughput, but on latency on trivial I/O
operations like expanding a share folder. Of course I would want to see
some graphs at end of test to show a relationship between the performance
metrics and numbers of simulated users.


For disk performance, try I/O Meter from www.iometer.org

It's the tool HP will get you to use if you claim any suspect disk
performance. I haven't used it myself, but it is scriptable.

Standard Benchmarks for File Servers
 

"Will" wrote in message
...
IOMeter is targeted at testing and benchmarking locally attached drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.


I have to disagree. I have used IOMeter to test SAN attached drives on
Hitatchi, StorageWorks, and EMC. Setup one worker process per CPU, what its
read/write policy is, how long you want it to run ... and yer in business.

I really like the tool, but it isnt going to benchmark how long it takes to
enumerate a folders contents, or things like that.

- LC

Standard Benchmarks for File Servers
 

"Nut Cracker" wrote in message
t...
"Will" wrote in message
...
IOMeter is targeted at testing and benchmarking locally attached drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.

I have to disagree. I have used IOMeter to test SAN attached drives on
Hitatchi, StorageWorks, and EMC. Setup one worker process per CPU, what
its
read/write policy is, how long you want it to run ... and yer in business.

Right, I should have said it is targeted at testing block-level performance
of locally or SAN-mounted drives.

I'm trying to test SMB NAS-type file shares.

--
Will

Standard Benchmarks for File Servers
 

"Will" wrote in message
...
"Nut Cracker" wrote in message
t...
"Will" wrote in message
...
IOMeter is targeted at testing and benchmarking locally attached
drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.

I have to disagree. I have used IOMeter to test SAN attached drives on
Hitatchi, StorageWorks, and EMC. Setup one worker process per CPU, what
its
read/write policy is, how long you want it to run ... and yer in
business.

Right, I should have said it is targeted at testing block-level
performance
of locally or SAN-mounted drives.


it tests by reading and writing to a file on a formatted drive .... so I
dont really think its a disk-block/sector-level tool.

- LC

Standard Benchmarks for File Servers
 

"Will" wrote in message
IOMeter is targeted at testing and benchmarking locally attached drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.

I reckon the only way to do that properly is to script typical behaviour
using a proper testing tool - something like WinRunner and try out the same
user scenario woth different HW setups and different user loads.

Big job, expensive tools, bloody good money if you can learn how to do it
properly.

Standard Benchmarks for File Servers
 

I found the old Ziff Davis tool named NetBench. It tests over SMB but it
seems pretty limited. It's creating some arbitrary benchmark number for
comparison of different hardware.

--
Will

"Jez T" wrote in message
...

"Will" wrote in message
IOMeter is targeted at testing and benchmarking locally attached drives.
I'm focused on multi-user performance of Microsoft file shares over the
network.

I reckon the only way to do that properly is to script typical behaviour
using a proper testing tool - something like WinRunner and try out the
same
user scenario woth different HW setups and different user loads.

Big job, expensive tools, bloody good money if you can learn how to do it
properly.