StoreOnce 4900 painfully slow replication

[Angyss]

Hello,

We have two StoreOnce 4900 which are connected by two fiber optic adapters, 10 GBe each, bonded as one connection. Some years ago when replication was first configured it used to replicate around 35 TB in about twelve hours, but now it's been over 15 days and it's barely 50% of 80TB.

We already checked network issues, the link is idle as it is almost entirely dedicated to this replication. configuration seems to be ok, no limits as for bandwidth or anything, maximum tapes to be replicated is 96, but throughput reaches around 9 MBps.

Testing with local VSAs bandwith behaves as 1Gbps, but between the 4900 it'd look like a 100 Mbps link. Firmware version is the latest that was recommended 3.16.5.

Any suggestions on what may be happening?

[NarayanKK-SO]

Hello,

The following issues could be one of the causes for slow replication

1. Storage space occupany on the source/target reaching the SpaceOver state.

2. Synchronisation issue between threads involved in a data job. Also if the source has open files. Replication will be paused until the control on the file is released.

3. Blackout windows set for replication.

The following steps can be performed to see if that impacts the performance

1. Schedule a downtime. Reboot the target first and then the source.

2. Clear the Coredump files and other unwanted stuff from the system and check for performance. Then must be done by HPE. A case can be logged with the fresh set of logs.

3. Disable blackout windows (if any) set for replication.

4. Download the Concepts and Configuration Guidelines for more info on Replication configuration.

Thanks