Start a Conversation

Unsolved

This post is more than 5 years old

G

7137

November 19th, 2015 02:00

MD3620i Remote Replication - Best Practise

Hi

I have some questions over the best way to configure remote replication on a pair of MD3620i arrays.

I have the replication working but my questions are more about the most supportable configuration.

1) Firstly the replication repositories are not displayed in the storage view in Storage Manager as far as I can see. I see that in the Fibre based replication the repositories appear in the view with the disk groups and virtual disks. Is there a way to display this in the iSCSI based systems ?

What seems to happen is the repository space gets assigned to a disk group, reducing the free space or the size of an existing virtual disk.

You can work it out be looking at the replication group but it's not obvious.

2) To make it easier to see where the replication repository is I have created a dedicate disk group called Replication Repository1 and manually created the repository in this group. So I have a virtual disk group of say 3TB which shows 2.5TB Free Capacity (assuming I create a 0.5TB repository).

This makes it easer to support. Any thoughts ?

3) Replication Groups - You cannot promote individual secondary replicated disks to primary. This seems to only be possible at the Replication Group level. Promoting a Replication Group will promote all the replicated virtual disks in the group.

I will probably only ever failover the entire system so a single Replication Group hold all the replicated disks would work. However for flexibility I would like the option to fail the virtual disks over separately.

So I would need to create 1 replication group per replicated virtual disk to be able to fail them over and back independently. Any issues with this ?

4) Replication Repository sizing - is 20% too high ? I have 32TB virtual disks, which will not consume 6TB of data in the 10 minutes between replication cycles.

Is there any reason I can't size these down to the match the expected volume of data I will acquire in each 10 minutes cycle (with some contingency allowance of course) ?

5) Any other gotchas I should be aware of ?

Many Thanks - Gary

No Responses!
No Events found!

Top