I have been trying to find out what the best practice for which datastores to present to a cluster.
The environment I am looking at has 10 clusters with about 10 ESXi hosts in each. There are also around 100 datastores that are currently presented to every host in every cluster.
To me this seems wrong as it means every ESXi host has 100 datastores, 90 of which are for data that belongs to other clusters, and I imagine a huge zoning config for the SAN switches.
I would think the best practice would be to setup the clusters so that only the datastores for that cluster are visible to the cluster. So in the environment I describe there would be something like 10 data stores presented to each cluster, and other clusters would not see datastores used in another cluster. This assumes that the purpose of the cluster is to group VM's that require similar availability and need access to the same datastores.
This is ultimately preparation for creating Host Profiles. I don't like the present setup as it would mean adding 100 datastores to a host profile which would work out to using the same Host Profile for all clusters.
I have been trying to find a best practice about assigning datastores to clusters and have not found anything related to this. If anyone has experience with this I would appreciate any comments.