VXRAIL and vSAN

[vmwarerati]

Hi All,

We are in process of evaluating a VXRAIL G Series Gen-4 (I guess that's vxrail 3.5 - released in June 2016?) - it's a hybrid configuration with 1 x 400GB SSD and 3 x 2TB per node in a 3 Node configuration to start with. I have the following questions:

a) Does the vSAN 6.2 design and sizing guide applies to VXRAIL - when designing /sizing?

b) VXRAIL G Series Gen-4 - does that comes pre-installed with vSphere 6u2 and vSAN 6.2?

c) I have read some articles where a 3 Node VSAN cluster is not always the recommended option and a 4 Node is preferred / recommended - is this the case with VXRAIL as well?

d) Calculating the usable capacity on VXRAIL (3 Nodes) with 1 x 400GB SSD and 3 x 2TB per node, that should reveal RAW 18TB (Base-10) capacity? with FTT=1 that will be 9TiB Base - 2 usable capacity - as that will make another copy of the data blocks across the cluster - Right?

e) I will be hosting about 50 - 70 Virtual machines on this VXRAIL, as per my understanding the FTT=1 is on a per disk group basis, If I create a single disk-group across the entire 3 node cluster - I will have all my virtual machines protected with FTT=1 for a single failure ( of a host / disk in the disk group)?

Will be grateful if someone can assist here - there's alot to consider when sizing the VSAN - even with vxrail - thanks

Best regards

Umar

[MBrownWFP]

I do not have direct experience with VXRAIL but my understanding is it is essentially a marketing name for Dell VSAN-Ready nodes that were designed and optimized specifically for VMware HCI.

a) As far as I know yes

b) Not sure this but should be easy to find an answer in VXRAIL documentation

c) Yes 4-node is recommended but this is not a hard limit, we run a 3-node cluster in our DR datacenter (HPE gear not VXRAIL) and it functions fine. Maintenance can be a little tricky as there are not enough members to maintain duplicate objects plus a witness if one node needs to be evacuated but VSAN can rebuild the mirror objects once full membership is restored. For a primary site I would not recommend as you lose redundancy temporarily (but not availability).

d) Yes, FTT=1 will duplicate all VSAN objects within the cluster resulting in 1/2 usable capacity compared to raw.

e) FTT is configured within a storage policy which is then applied at VSAN datastore level (a second policy with different FTT can be created and set for individual VMs if required). You will have a single VSAN datastore the spans the cluster, this is where your VMs will live. Disk groups are created at the host level. Given your disk layout you will have one disk group per host comprised of your single 400GB SSD plus 3x 2TB HDD.

Hope that helps.

Matt

[aleksey]

Matt is right on the money for all of the answers. I work closely with VXRAIL engineering teams, so I can confirm some of the answers.

re: A and B are also firm YES.

re: C. 3 node configurations aren't supported with VXRAIL 3.5 release, but will be supported with 4.0 (look for announcements from a couple of weeks ago at Dell World)

[IvanATDell]

c) 3 node cluster is now supported on VXRail 4.0. Here is the link to official announcement: Dell EMC VxRail 4.0 Announcement – New Models, More Use Cases - Virtual Blocks

[BAURCHK]

HI

We are evaluating to deploy almost the same configuration and spec.

Is there any issues you have experienced so far that you could share with us ?

Thanks.

[depping]

if there were issues, than they were more than likely software based, and I would assume that those have been solved by now.VxRail and vSAN has come a long way since 2016....

[srodenburg]

Can confirm VxRail 4.x working fine.

Would stay away from 3-nodes configs though (for all the reason already mentioned). Rather go with 4 smaller nodes than 3 bigger ones. Makes daily admin life better 😉