Is it possible to setup Asynchronous volume replication between a master and one or more slave nodes?
I have a situation where we have an existing backup system and we are using open-e to host a 4TB NFS export, which we use as a target for an rsync based backup and versioning system, very similar to rsnapshot. We use hardlinking for our versioning system, as not to duplicate files on disk.
However when it comes to replication using file based replication within DSS I believe that we will lose the hardlinking? Which would mean around a x10 increase in storage requirements at the destination, and ~ x10,000 or so increase in network traffic. Am I wrong?
So this leads me to believe that our only option is volume replication, which is fine, however we need it to ideally be asynchronous, and to be able to have two slave nodes if possible. Both slave nodes being on the end of wan links. One of the wan links being very good, about 30mpbs at a distance of 1km. and the other wan link being very bad, about 2-4mbps on the other side of the world.
The volume replication has only Synchronous and not Asynchronous, we are working on this but no set ETA as of yet due to our current task list that is already scheduled for this year but engineers are wanting to have this for the Volume Replication.
In my test setup, which will go in production at the end of this week, I setup the (synchronous) volume replication task with a schedule:
Starting at 7 pm, stopping at 7 am.
This is then almost like asynchronous at working times, and synchronous at non-working hours. What do you think about this solution?
The DSS V6 is synchronous only for volume replication and there is no set stop and start values - only asynchronous works like this and we dont have this ready.
prahn....
I think you'll find once you update(without switching to sync rep), you will loose the ability to manage this schedule(this was removed). And the task will need to be recreated to work properly.
Switch back to synchronous, and update. Performance will be better.