Ever since the announcement of vSphere 6 this week, one of the coolest things I have noticed is the improvements of VSAN 2.0. One one hand, if you know how to configure VSANs in 5.5U1, you will have no problem configuring VSAN 6. On the other hand, there are significant differences below the hood.
Here is a list of some of the main differences:
1. Brand New File System for VSANs 2.0
The original VMFS-L (local) has been replaced with VSAN-FS/Virsto FS. Virsto is a company that was acquired by VMware a couple of years back. Make sure to upgrade the file system itself (a manual process) if upgrading from 5.5 U1 to 6.
2. Fault/Failure Domains
Fault/Failure domains allow you to group multiple esxi hosts into a single domain, treating them as a unit. This is very useful if multiple servers are connected to a single rack and you fear that the loss of power to a rack could affect multiple esxi hosts. Fault domains can be created very easily with the Web Client. This does NOT mean you can create a metro cluster.
3. 100% All-Flash-based VSANs
This feature is finally supported. However, all SSD drives will be initially seen as SSD (and not capacity drives). You can use the esxcli command to indicate which drives will be used for capacity instead of caching. When usgin an All-Flash VSAN, the ssd drives NOT used for capacity are only used for write-buffering. The All-Flash VSAN comes at a higher price though
1. Find the disks to mark as capacity disks with esxcli storage core device list
2. Mark them with esxcli vsan storage tag add -t capacityFlash -d device_name
3. Verify the previous step with the command vdq -q -d device_name. Look for the parameter
isCapacityFlash = 1
VSANs 6 support 64 hosts instead of 32. Vmdks can now be 62tbs instead of 2tbs. VMware now supports 200 vms per host (instead of 100). You can now get 40k iops per host (with a hybrid vsan that is). More than twice as much with an all-flash vsan. Lastly, now you can have 9000 components per host instead of 3000. There is also support for directly-attached JBODs. Lastly, you can now see more information with the Web Client, including re-syncing progress.
5. New commands available
Log into the esxi host and type esxcli vsan. You will notice new commands (for example the ones for fault domains). The RVC also has a command to manually rebalance (vsan.proactive_rebalance). Rebalancing happens automatically if the drives go above 80% of space utilization. This command allows rebalancing at any time, which can be useful when adding new hosts with new disks.