cancel
Showing results for 
Search instead for 
Did you mean: 

Could we consider VVR as mirorred volume?

Altimate1
Level 6
Partner Accredited

Hi,

Let's consider I want to have one site to replicate data to a second site.

I could probably setup SFW+VVR on both sites and configure VVR so that to be sure data volume at source location is replicated to remote site.

The main question is:

If disk at primary site become unavailable, does SF try to serve READs/WRITEs from that main host from/to the remote location (as it would do when configured as mirorred disks?)

I don't found obvious answer in the documentation about a such feature?

Regards

1 ACCEPTED SOLUTION

Accepted Solutions

Wally_Heim
Level 6
Employee

Hi Altimate1,

The simple answer is no.  VVR does not work like a mirror in that it will not service I/O requests from the secondary if the primary volumes/disks become unavailable.

However, if you through in HA and GCO, you can get SFW-HA to automatically failover your application to the secondary site and use the data at that location while you repair the failure at the original primary site.

Thanks,

Wally

View solution in original post

5 REPLIES 5

Gaurav_S
Moderator
Moderator
   VIP    Certified

Hello,

I am certain this thing won't be documented anywhere in public documentation however thinking about the functionality on how it works, VVR won't try to serve read/writes from secondary ... the problem here is ownership of the volumes....

by design in VVR, filesystems can be mounted only on primary site, on secondary , you can only have diskgroup imported & volumes started but if replication is active, it won';t allow you to mount the filesystem.

So I believe, there is a difference in VVR & mirrored disk, VVR doesn't solve this purpose of what you are looking for..

 

Gaurav

TonyGriffiths
Level 6
Employee Accredited Certified

Hi

Just to add to Gaurav's reponse

If you have a redundant volumes (mirror etc) at the primary  and a disk is compromised, then i/os at the primary will continue to be servive by the primary volumes.

If you do not have redundant volumes and a disk is compromised, then your primary data volumes are also compromised. You will still have replicated data at the secondary which you can make available to the secondary server.

Wally_Heim
Level 6
Employee

Hi Altimate1,

The simple answer is no.  VVR does not work like a mirror in that it will not service I/O requests from the secondary if the primary volumes/disks become unavailable.

However, if you through in HA and GCO, you can get SFW-HA to automatically failover your application to the secondary site and use the data at that location while you repair the failure at the original primary site.

Thanks,

Wally

mikebounds
Level 6
Partner Accredited

How far apart are your sites, if they are too far apart for sync replication, then if VVR were to work like you want (which is doesn't - as quite simply put, regardless of sync/async mode, a write MUST complete to the primary SRL which must be SAN attached) then your application would slow down dramatcially to the point where it probably wouldn't function.  If your sites are close together so you could do sync replication, then if you don't have fibre channel between sites then you might be able to use SCSI over IP so that you could use VM mirroring.

Mike

Altimate1
Level 6
Partner Accredited

Thanks alls,

I have now a better understanding of what we can do with VVR.

The most useful answer was thoses clearly stating NO, VVR CAN't OFFER THIS.
After reviewing project context, the solution is apparently to use volume manager to do Raid 1 on local and remote disk (the latency was acceptable as distance is more close campus than long distance.

This way on if the disk storage on the primary site goes down or fail I/O can survive throught partner (remote disk).

For site failover we will sure have to Import the volume so that to be able to work (R/W) from the recovery site.

Thank you again.
Regards