Protect Storage Pool Performance

Jackal

Active Newcomer
Joined
May 16, 2018
Messages
7
Reaction score
0
Points
0
PREDATAR Control23

We have a primary and secondary Spectrum Protect servers based on IBM's large blueprint. Our daily protected amount is around 16 TB with 2TB of written data, the rest is deduplicated or compressed. Under normal conditions, the protect storage pool process would take 60 - 90 minutes to complete.

A week ago, while experimenting with the TCPWINDOWSIZE on the server side, we stopped the protect storage pool process and have not been able to get back to our normal results; even after reversing all of our changes. We do have an active PMR open with IBM at the moment to look into the issue but would like to know if anyone here has any ideas.
 
PREDATAR Control23

Hi Jackal,

Are you able to tell us what version of Spectrum Protect server/s you're using between the Protect STG process and the OS's involved please?

I and another forum member have had cases open with IBM related to the 'Protect Stg Pool' process with it failing on a small number of extents and thus leaving orphaned sessions open on the destination server.
Approx. 152GB in 3 hours over a 1 GB WAN VPN link via SSL [SP 8.1.4.012 - WIN > Linux]


Thanks
 
PREDATAR Control23

Spectrum Protect: 8.1.4.014
OS: AIX 7.1.4 TL 4 (both sides)
WAN: 4 x 10 Gbps MLAG

I ran through this link yesterday and pulled the numbers:

http://www-01.ibm.com/support/docview.wss?uid=swg1IT14233

Results from our system:
Source: 1519117853
Target: 1675270701

I wouldn't think there would be orphaned sessions since we have to stop the instance to update the window size.

DB2 team is looking into a rouge DB2 thread at the moment but we haven't heard anything yet.
 
PREDATAR Control23

Here is today's progress:
Extents protected: 8997306 of 27008718. Extents failed
to protect: 0. Extents deleted: 7657099 of
7657099. Amount protected: 8,124 MB of 1,721
GB. Amount failed: 0 bytes. Amount transferred:
8,216 MB. Elapsed time: 0 Days, 3 Hours, 10 Minutes.

It usually took a while to start but once it did, it would move a lot of data quickly. This is a 'normal' day, green is replication traffic to the other server.

1529509639157.png
 
PREDATAR Control23

Hi,

The obvious settings has been set?

On both sides on AIX

no -o rfc1323=1


And

tcpwindowsize 0

in dsmserv.opt on both sides
 
PREDATAR Control23

We have TCPWINDOWSIZE 128 now, I have thought about using 0.

Once it gets out if its own way it takes off!


Extents
protected: 12342181 of 30023638. Extents failed
to protect: 0. Extents deleted: 7657099 of
7657099. Amount protected: 33,851 MB of 2,124
GB. Amount failed: 0 bytes. Amount transferred:
33,942 MB. Elapsed time: 0 Days, 6 Hours, 29
Minutes.

Extents
protected: 14982176 of 32587819. Extents failed
to protect: 0. Extents deleted: 7657099 of
7657099. Amount protected: 225 GB of 2,457 GB.
Amount failed: 0 bytes. Amount transferred: 226
GB. Elapsed time: 0 Days, 7 Hours, 32 Minutes.

Extents
protected: 23728626 of 35295058. Extents failed
to protect: 0. Extents deleted: 7657099 of
7657099. Amount protected: 1,284 GB of 2,812
GB. Amount failed: 0 bytes. Amount transferred:
1,280 GB. Elapsed time: 0 Days, 8 Hours, 14
Minutes.

Extents
protected: 28365616 of 35295058. Extents failed
to protect: 0. Extents deleted: 7657099 of
7657099. Amount protected: 1,908 GB of 2,812
GB. Amount failed: 0 bytes. Amount transferred:
1,901 GB. Elapsed time: 0 Days, 8 Hours, 26
Minutes.
 
Top