0

FCI is said to be an HA solution.

If one node crashes, it spins up the 2nd node so why is it said to not be a DR solution? There was a disaster at node 1 and node 2 picked up from there. Isn't that isaster recovery?

Further lets assume that the disk is replicated, then can we say that it is a DR solution? Here if node or disk crashes, we have the other node or disk to take over. Isn't this disaster recovery?

Improve this question
7
  • "If one node crashes", it is not a disaster. A disaster is when your entire data centre, or the whole city with all three redundant data centres, is flooded or blown up by a supervolcano.
    – mustaccio
    May 11, 2021 at 15:39
  • Ok in that case assume node 1 is located in USA and node 2 is located in Europe. There is flood in USA and data centre has collapsed. FCI will failover to the Europe data centre node. So why is FCI not a DR solution here?
    – variable
    May 11, 2021 at 15:46
  • It could be, if you could make it (WSFC) work with across-the-pond latencies and such. Since the risk of a disaster is much lower than the risk of hardware failures, options that are cheaper and easier to maintain are often chosen for DR.
    – mustaccio
    May 11, 2021 at 15:50
  • 1
    An FCI with asyncronous disk replication in a stretch cluster can protect against datacenter loss, but that is a complex solution and requires a sophisticated SAN solution to manage. Out-of-the box and in 99% of deployments an FCI is a local HA solution, and you augment that with some remote, async copy, through backups, an AG or disk replication.
    – David Browne - Microsoft
    May 11, 2021 at 17:53
  • 1
    HA is all about maintaining services to your data. DR is all about getting your data back should you lose it. A traditional FCI has two servers but one shared disk resource and since the data is a single resource, should you lose that disk resource, it doesn't matter how many servers are available. An AG basically repurposes the FCI architecture to allow each node to manage its own disk resources so now you have both server availability and data redundancy.
    – Queue Mann
    May 11, 2021 at 18:09

0

Reset to default