Backup copy (HPE StoreOnce Repository) performance

[jobhlidal]

Model: StoreOnce 5250
We use new feature of Veeam 10 to perform Catalyst replica from primary catalyst repository to secondary catalyst repository.
Backup job to primary catalyst repository uses LAN connection from gateway server to catalyst store.
Catalyst replica uses FC gateway connection to secondary catalyst repository.
So replication uses 16Gbit FC between storeonce boxes.

The first replication transferred 46TB of user data by the speed about 1TB /hour. I could see from the storeonce report that it used max 40 connections for the replication.
I would like to know if it is possible to control the number of connections which are used during the replication? Or somehow to increase the performance/speed of the replication?

We are going to replicate catalyst stores with more than 300TB of user data which in current speed means about 12 days.
Thank you
Jan

[HannesK]

Hello,
sorry that I do not understand your setup. But I would like to verify it, because our use guide says, that the same connection type is required.

Question 1: your setup is this?

Code: Select all

(Proxy -> Ethernet ->) Gateway server source -> Ethernet -> StoreOnce1 -> fibre channel -> StoreOnce2 ?

Question 2: StoreOnce1 and StoreOnce2 are configured with the same gateway server?

Catalyst replica uses FC gateway connection to secondary catalyst repository.

Question 3: which gateway do you mean in this case? Catalyst copy is direct communication between two StoreOnce systems. There is no Veeam component involved in the copy process.

Question 4: What did HPE support say on that question? Veeam just leverages 3rd party technology. So chances are high at the vendor of the technology.

Best regards,
Hannes

[jobhlidal]

Hello Hannes.
Q1:
My setup is:
Repositories configured in Veeam
Repository1
Proxy/Gateway server source => LAN => StoreOnce1 (Catalyst Store 1)
Repository2
Proxy/Gateway server target => FibreChannel => StoreOnce2 (Catlyst Store 2)
Connection between StoreOnce1 and StoreOnce 2 is Fibre Channel
So in my setup, target HPE StoreOnce repository configured over Fibre Channel and both StoreOnce boxes are also connected over Fibre Channel.

Q2: No. Each StoreOnce uses different gateway server.

Q3: I know that Catalyst copy does not use any Veeam component to transfer data. But the connection which will be used for direct communication between StoreOnce boxes is based on configuration of target catalyst repository in Veeam.

This direct connection must be of the same type as the connection that you select when adding the target HPE StoreOnce. For example, if you connected the target HPE StoreOnce repository over Fibre Channel, you must connect the source HPE StoreOnce to the target HPE StoreOnce over Fibre Channel.

In my case the catalyst copy goes over Fibre Channel which is correct. But the performance is not good. So I am looking for same information if it is possible to tune this performance on Veeam side. Or some recommendation how to get the best performance for catalyst copy.

Q4: I do not have opened case in HPE.

[HannesK]

Hello,
then I have only one guess left: how many task slots to you have configured on the StoreOnce repositories? Can you increase them?

Best regards,
Hannes

[jobhlidal]

Hello.

We configure Catalyst repositories with "No limit" for number of connections.
But as you said the replication process should not use any Veeam component, so I would expect that number of connection configured for Veeam repository should not impact catalyst replication because Veeam is not involved in replication.

Best regards
Jan

[HannesK]

Hello Jan,
I expected the same like you.

But developers told me, that the task setting also limits the number of concurrent copied files (to avoid overload of the systems)

With unlimited being set, I can only refer to HPE support. If you have a chance to tell us the outcome, that would be nice.

Best regards,
Hannes

[foggy]

Parallel jobs should allow to parallelize the workload and involve more storage connections.