1 Reply Latest reply: Apr 20, 2012 2:30 PM by dave RSS

Performance SPARC T5120 / Networker / DataDomain / Boost

B_Beat

Hi

 

We currently work on implementing our new DataDomain in the networker environment. The networker Server is a T5120 running Solaris 10, with SAN attached storage (EMC VMax) and a DataDomain DD670 connected to 1GB Lan device. On top of it we run Networker 7.6.2 with Boost Option enabled.

 

To explain the problem I should start with a Configuration like this:

 

10 SAN Filesystems total capacity 1.7TB

1 DD-Boost Device (10 Sessions)

1 SaveGrp

1 Pool

 

Result -->     Average throughput 38MBytes/sec running für 12.5hours (after 3 runs)

 

Then we tried to add a 2. DD-Device

 

10 SAN Filesystems total capacity 1.7TB

2 DD-Boost Device (10 Sessions/each)

1 SaveGrp

1 Pool

 

Result -->     Average throughput 36MBytes/sec per Device running für 6.5hours (after 3 runs)

 

More devices

 

10 SAN Filesystems total capacity 1.7TB

4 DD-Boost Device (6 Sessions/each)

1 SaveGrp

1 Pool

 

Result -->     Average throughput 28MBytes/sec per Device running für 4.25hours (after 3 runs)

 

and more devices

 

10 SAN Filesystems total capacity 1.7TB

8 DD-Boost Device (4 Sessions/each)

1 SaveGrp

1 Pool

 

Result -->     Average throughput 19MBytes/sec per Device running für 3.1hours (after 3 runs)

 

You see, the throughput raised from 36 to 152MByte/sec when defining more devices, but to save one single filesystem with 500GB will still take too much time when we can't break this 36Mbyte/sec for a single device.

 

Is anyone here experiences similar problems/behavior? Are there any parameters to be checked?

Any ideas are welcome

 

Regards

Beat