Hi everyone , i'm working in a new Nutanix deployment and i'm into a dilema.
I need to migrate a storage cluster into Nutanix solution. Stogate cluster size is 18tb and i have 7 3060-G5 with 800gbs sd and 2tb hdd. Since customer is used to storage cluster availability i'm thining to deploy a Microsoft cluster with 18tb volume group attach but i'm not sure how data locality works in that case.
Another option is to create 5 4tb DFS servers without replication since i don't have enough space to do that . Problem with this option is that if one server goes down i'll have an outage.
Since i'm writing i'm thinking in a 3rd option but is may be too complex
5 Microsoft cluster of 4tb using volume groups to share storage and join all those cluster into a signle DFS namespace.
What do you think ? Would be greate a second point of view
Assuming that the customer requires the availability provided by a microsoft storage cluster, (I'm assuming for windows file shares, since the other options are talking about DFS), you can run that on top of Nutanix with 2 VM's (just like two physical servers previously), and use volume groups with in-guest iSCSI to make the cluster.
That should work just fine.
Given that the storage cluster is quite a bit larger than the amount of storage you have on any single one of your nodes, data locality isn't really a factor here since the VM is so large. We will keep those storage blocks as close to the VM as we can, but obviously we wouldn't not be able to keep all 18TB on a single 3060 node.
Also, I would recommend that you break that 18TB up into multiple storage volumes if you can.
Jon Kohler | Principal Architect, Nutanix | Nutanix NPX #003, VCDX #116 | @JonKohler Please Kudos if useful!
Finally i'm goinng to create a cluster with 2 vm and VG though ISCSI. I was really worrie about data locality but for File server workload is not really a concern regarding other customer expiriences. Obviously is a desired but not a must