#SM-BC

1 messages · Page 1 of 1 (latest)

distant yarrow
#

Can we run an SM-BC relation between two controller pairs in the same Cluster? The scenario is to be able to survive non-disruptively to a double controller outage in a 4 controller Cluster, serving block services to a vmware farm.

echo rune
#

So... 2 things here -

  1. I don't believe you can run SMBC on a 4 node cluster.

What are the supported platforms for SnapMirror business continuity? -
SMBC supports only two-node HA clusters, either AFF or ASA models.

  1. A 4 node cluster is intresting, If the pair that is holding Quorum/Epislon goes down on a 4 node the remaining 2 cannot elect a new master. and the remaining nodes stop serving to protect itself.

https://docs.netapp.com/us-en/ontap/system-admin/quorum-epsilon-concept.html

#

6 node and higher it's not a problem -

#

granted in 10+ years using and deploying ONTAP i've not seen or heard of that actually happening in the field.

tardy dune
#

If there's only the SAN workload for VMware I don't see a real benefit in your theoretical scenario (4-node cluster) vs. simply using 2x 2-node clusters. You would be replicating via SM-BC so you don't need the possibility to vol move.

distant yarrow
#

The usefulness of this 4 node Cluster has raised some question before. Actually, if they had the space to move everything to the first pair, transforming it into 2 x 2-node would have happened already. As for the probability of being it by a double controller failure, in 20 years working with Netapp I've not seen it too but, this particular Customer has seen two controller replacements in one year so, his perception might be affected. There is also the Health Care contents. I believe we all agree that the idea of someone waiting for medical records because of an outage that could have been prevented with a simple feature, is something that makes sense even if it seems a farfetched scenario.

#

..and thanks for sharing your opinion.

hollow light
#

If it is for healthcare/safety of life systems requiring 100% uptime, Metrocluster is what they should be looking at

#

Not SMBC

distant yarrow