I have been playing with SQL Server Clustering (FCI) setup with a Widnows Server 2016 Preview 4 and Windows Storage Replica.
This involves local storage and Cluster Shared Volume with Replication Enabled.
This is incomplete – if Node 1 goes offline the SQL Server role fails!
Windows Cluster Shared Volumes are a disk volume which acts as a clustered filesystem i.e. read-write on all nodes.
In fact, in Disk Management on the cold side of the cluster,the Volume is tagged with ‘(CSVFS)’
SMB 3.0 is used behind the scenes
How do I setup a Cluster Shared Volume?
What if a cluster node goes down?
No storage failover is needed when a node goes down
- sql server has no cluster dependency on the disks anymore!
There is still a co-ordinator node for writes.
What if all storage paths to a node go down?
If this is the co-ordinator node, another co-ordinator node is chosen and i/o rediection occurs. See http://www.smooth1.co.uk/mssql/mssql_csv_iorediection.html
How do Cluster Shared Volumes work internally?
Cluster Shared Volume Diagnostics
Cluster Shared Volume Performance Counters
Cluster Shared Volume Failure Handling