currently I'm designing a vSAN solution and found some things that are unclear to me.
One VM will leverage vSphere Fault Tolerance in this cluster, which is supported but still leaves some questions for me:
- vSAN All-Flash & FT both require 10 Gbit/s networking
-> Can I create a LAG on the dvSwitch with both 10 Gbit/s Uplinks or do I need a dedicated NIC for FT?
I'm well aware that vSAN can share uplinks with services like vMotion, VM Traffic, or Management, but I never saw a cluster with vSAN & FT before.
If sharing the uplinks is ok, do I need to set shares within the NIOC on the vDS?
If I use shares in NIOC, should vSAN be set to "high" and FT to "normal"?
Moreover, I'm planning a second LAG for vMotion, VM Traffic and Management.
This would also be 10 Gbit/s and reside on a vDS as well.
Would it make sense to put vSAN on the second LAG with shares configured, so that FT has dedicated bandwith?
Another idea is a LAG containing all 4 10 GbE NICs, with either shares or traffic shaping (even though shares are recommended instead of maximums)
The last thing that comes to my mind is consolidating the 2 dual-port nics and just use one dual-port NIC with 25 Gbit/s.
In that case I'd create a LAG with 2x 25 Gbit/s, put FT, vMotion, vSAN, Management and VM traffic on that LAG and set the shares.
I'm thankful for every opinion on this one.
Second question regarding the faul tolerance.
vSphere Fault Tolerance will create a secondary VM that resides on a different host than the primary VM.
Furthermore, the second VM should reside in a different datastore - thats where my problem starts.
In the vSAN Cluster I only have a single datastore, therefore I cannot place the secondary VM on a different datastore.
How do I handle that in a vSAN cluster?
The Cluster is a single site with 4-5 ESXi Hosts.
The VM would be provisioned with FTM-RAID1 and FTT-1.