Hello everyone, is there a way to achieve failover of a CIFS share with RPO of 0 and a small RTO between two datacenters without Metrocluster? Snapmirror Active Sync does not seems to support CIFS and Snapmirror Sync does not support SVM-DR.
#CIFS shares and RPO 0
wispy nexus
eager cairn
I never got it to work, but I've tried the same thing. You can configure volume snapmirrors with a synchronous policy configured, but when you create a CIFS share to the replicated volume, you're not actually seeing the active filesystem. Instead it shares out the most recent common snapshot copy... which I think can be configured to as frequently as every 30 minutes. But when you create a file on the source, you still won't see it on the destination share until a new update runs even though supposedly the AFS is synchronously replicated in the background. 🤷♂️
daring terrace
ONTAP Select with MetroCluster SDS might be an option
desert sphinx
Metrocluster is really the answer - can I ask why you are looking at alternatives?
wispy nexus
Currently we have one AFF cluster per site and are using software based replication in order to maintain low RPO (sql server always-on for databases, Hyper-V replica for VM, and Windows server Storage Replica for file shares). We had some issues with Storage Replica and wanted to migrate the shares directly on Ontap when we renew the platform.
We wanted to avoid Metrocluster because there seemed to be constraints in terms of network equipment to link the 2 sites. I was looking into Snapmirror-activesync technology but the documentation on the subject seems sparse.
daring terrace
well, if your network link between the sites does not meet the requirements for MetroCluster, then running SM active sync over it will not be a very fun experience either. You really need low latency and a reliable connection, no matter the replication technology
desert sphinx
Yeah, RPO zero should mean you don’t acknowledge writes until the remote site has a copy. Anything else isn’t really RPO 0