ONTAP Discussions

SnapMirror Issue

nindermaan
3,858 Views

Hi,

In my environment we have mirrored some volumes to DR site. As we have network bandwidth of 23mbps. But one urgent snapmirror is not going more than 10mbps.

We have tried the throttle value to 0 and 20000 also. still its not going more than 10mbps.

Regards,

Narender Mann

Storage Admin

7 REPLIES 7

nindermaan
3,858 Views

My Current Configuration is:

DC_Filer> options replication
replication.logical.reserved_transfers 0          (value might be overwritten in takeover)
replication.logical.transfer_limits current    (value might be overwritten in takeover)
replication.throttle.enable  on
replication.throttle.incoming.max_kbs 20000
replication.throttle.outgoing.max_kbs unlimited
replication.volume.reserved_transfers 0          (value might be overwritten in takeover)
replication.volume.transfer_limits current    (value might be overwritten in takeover)
replication.volume.use_auto_resync off        (value might be overwritten in takeover)

----------------------------------------------------------------------------------------------------------------------------


DR_Filer> options replication
replication.logical.reserved_transfers 0          (value might be overwritten in takeover)
replication.logical.transfer_limits current    (value might be overwritten in takeover)
replication.throttle.enable  on
replication.throttle.incoming.max_kbs unlimited
replication.throttle.outgoing.max_kbs 20000
replication.volume.reserved_transfers 0          (value might be overwritten in takeover)
replication.volume.transfer_limits current    (value might be overwritten in takeover)
replication.volume.use_auto_resync off        (value might be overwritten in takeover)
DR_Filer> rdfile /etc/snapmirror.conf
#Regenerated by registry Fri Jun 24 05:15:39 GMT 2011
DC_Filer:Vol0_Test DR_Filer:Vol0_Test - 15,45 * * *

DR_Filer>

scottgelb
3,858 Views

It could also just be as fast as the source controller/aggr/volume can send... you can check disk utilization with statit or perfstat for more details.  You can also send snapmirror to a null device.. "snapmirror store volname null" (or if ontap 8, "smtape backup volname null"  and see how fast it can mirror without the wan.  Just make sure to delete the orphaned snapshot after so you don't have it take space later from that snapmirror to tape..

nindermaan
3,858 Views

Hi Scott,

I have tried the same commands but there's no performance bottleneck. And sending snapmirror to null device means.?

scottgelb
3,858 Views

Null is a bit bucket.. It should show how fast snapmirror can read from the source volume eliminating network and target which could be (or not) the bottleneck.

Typos Sent on Blackberry Wireless

eric_barlier
3,858 Views

Your best bet is to try what Scott suggested, have you tried it yet?

Also, that specific volume you have "problems" with, does it contain a very high number of inodes?

Does it reside in a different aggregate than the other volumes?

Does this volume only do 10mb even if you run the update in solo?

Cheers,

Eric

ademuynck
3,858 Views

Hi,

Have you check your network config ? Maybe, you have a qos policy on your WAN ?

Check your WAN router, on both side (DC_Filer and DR_Filer)...

Hope it help you

nindermaan
3,858 Views

Hi Ademuynck,

I checked out our network config on both the ends. When multiple sessions starts from the same filer it consumes the n\w bandwidth to 20mbps.

but if only single session comes that its not using more than 10 mbps.

Public