Syncsort and DR

[Willybobs27]

I have a customer who has Primary EMC storage and wants to use a DR service based on NetApp and VMware.

A good solution seems to be Syncsort and the instant Virtualisation feature in a DR Scenario.

However I have a few questions:

There will be no NetApp on the primary site so all Backups will go over the WAN to the DR Site, is this supported?

If DR is initiated using instant recovery and the VMs are up and operational, how do we restore back to the EMC primary site? Is this even feasible without a local NetApp?

[radek_kubka]

There will be no NetApp on the primary site so all Backups will go over the WAN to the DR Site, is this supported?

yes, that's fine if the pipe is OK to transfer block level changes.

If DR is initiated using instant recovery and the VMs are up and operational, how do we restore back to the EMC primary site?

Instant Virtualization maps backed up LUNs from snapshots as RDMs over iSCSI to a newly created VM - so you can't truly restore it back, as there are no means of resynchronization of the changes.

 Is this even feasible without a local NetApp?

It is, but having local NetApp filer makes restores way quicker.

[Willybobs27]

Thanks for that.

One other question, if I have Tape Integration is the tape directly attached to the NetApp or does it need a server to mount the snapshot and then write the snapshot contents to tape?

[radek_kubka]

Syncsort uses NDMP for tape backups, so directly attached tape is fine.

[UPADHYAYULA]

If DR is initiated using instant recovery and the VMs are up and operational, how do we restore back to the EMC primary site?
Instant Virtualization maps backed up LUNs from snapshots as RDMs over iSCSI to a newly created VM - so you can't truly restore it back, as there are no means of resynchronization of the changes.

In a DR scenario, it is possible to restore back to the primary site after using Instant Virtualization (IV) by using the Rapid Return to Production (RRP) feature. This feature (invoked by a simple wizard) allows to seamlessly convert the instantly recovered VM into a production VM while the users are continuously accessing the machine (uses VMware Storage vMotion).

PS: I work for Syncsort and would be happy to provide further details on how to accomplish this.

[radek_kubka]

Thanks for pointing that out.

However, RRP may be not feasible when DR site is 'truly' remote & connected to the production site via high latency / low bandwidth link. The mechanism (basically copying across the entire VM) is substantially different from a SnapMirror resynchronisation, which sends only block-level incremental changes.