VMware Cloud Community
9990374530
Enthusiast
Enthusiast
Jump to solution

VSAN expansion

Hello experts,

I would would like to see the possibility of mounting a VSAN of about 120 Tb or so in our each virtual infra Datacenter?
As we have 9 Dell hosts with their local disks and this is giving us about 31.44 TB in each CPD.

1. Do we need to shutdown the ESXI host that we have to expand the VSAN space in each Datacenter to have about 120 tb approx. net in each CPD?
2. Can we change those disks for larger disks, if additional disks can be plugged into the slots of the chassis that we have empty, etc?

3. How can I ensure to fulfill the requirement 120TB, how much drives I will have to buy?

 

 

Regards

PK

 

Reply
0 Kudos
1 Solution

Accepted Solutions
mmangold
Enthusiast
Enthusiast
Jump to solution

I believe the answer to your specific questions are as follows:

1. Yes. I would evacuate the first host, and Disconnect it from the vSAN, migrating its VMs across the remaining hosts. Next, you will populate the new SSDs into the ESXi host- after which, you will want to destroy the existing disk-groups, followed by re-creating them using the new numbers (e.g. from 1x-cache w/3x-capacity to 1x-cache w/5x-capacity.) Note that there should be a 1:1 ratio of cache drives to disk groups- that is, for each disk-group, there will be 1x cache SSD and some number of capacity disks, where the sum of the capacity disks total the desired 120Tb.

2. Yes. Populate each ~120TB VMHost with the calculated combination cache- and capacity-drives (aka disk-group) drives. A cache disk should be at least ~10% the size of the disks used for capacity (for hybrid vSANs; note that there is no 10% cache:capacity ratio requirement for all-flash array vSANs.

3. (Total space desired) / ((total number of drive bays in VMHost) - (number of cache disks per host)) = capacity-disk size required. (e.g.: if each server has 12 drive bays and there are to be two disk groups, then 10 drive bays remain, across which our total storage space is to be distributed. Therefore, 120Tb / 10 disks = 12 Tb. If there are three (minimum required) VMHosts at each DC, then 12/3 = 4Tb per capacity disk per VMHost (and for a 4Tb capacity disk, a 500Gb cache disk would suffice- but just do the all-flash, if at all possible: you'll be happy you did!)

Once the host is fully re-populated and the disk groups have been re-created, next re-join the ESXi host to the cluster and migrate back the VMs, so that the process may be reiterated on the remaining hosts in the environment.

Please review: https://blogs.vmware.com/virtualblocks/2018/08/23/extending-all-flash-vsan-cache-tier-sizing-require... and https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.vsan-planning.doc/GUID-1D6AD25A-459...for additional information.

Good luck!

View solution in original post

6 Replies
mmangold
Enthusiast
Enthusiast
Jump to solution

I believe the answer to your specific questions are as follows:

1. Yes. I would evacuate the first host, and Disconnect it from the vSAN, migrating its VMs across the remaining hosts. Next, you will populate the new SSDs into the ESXi host- after which, you will want to destroy the existing disk-groups, followed by re-creating them using the new numbers (e.g. from 1x-cache w/3x-capacity to 1x-cache w/5x-capacity.) Note that there should be a 1:1 ratio of cache drives to disk groups- that is, for each disk-group, there will be 1x cache SSD and some number of capacity disks, where the sum of the capacity disks total the desired 120Tb.

2. Yes. Populate each ~120TB VMHost with the calculated combination cache- and capacity-drives (aka disk-group) drives. A cache disk should be at least ~10% the size of the disks used for capacity (for hybrid vSANs; note that there is no 10% cache:capacity ratio requirement for all-flash array vSANs.

3. (Total space desired) / ((total number of drive bays in VMHost) - (number of cache disks per host)) = capacity-disk size required. (e.g.: if each server has 12 drive bays and there are to be two disk groups, then 10 drive bays remain, across which our total storage space is to be distributed. Therefore, 120Tb / 10 disks = 12 Tb. If there are three (minimum required) VMHosts at each DC, then 12/3 = 4Tb per capacity disk per VMHost (and for a 4Tb capacity disk, a 500Gb cache disk would suffice- but just do the all-flash, if at all possible: you'll be happy you did!)

Once the host is fully re-populated and the disk groups have been re-created, next re-join the ESXi host to the cluster and migrate back the VMs, so that the process may be reiterated on the remaining hosts in the environment.

Please review: https://blogs.vmware.com/virtualblocks/2018/08/23/extending-all-flash-vsan-cache-tier-sizing-require... and https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.vsan-planning.doc/GUID-1D6AD25A-459...for additional information.

Good luck!

9990374530
Enthusiast
Enthusiast
Jump to solution

Hello,

Thanks for your response.

So, do we also need to consider FTT, RF and RAID factor in to consideration while expanding existing vSAN Cluster?

 

 

 

Reply
0 Kudos
RajeevVCP4
Expert
Expert
Jump to solution

If you have login id you can use vsan sizer tool

 

https://vsansizer.esp.vmware.com/login?returnUrl=%2Fhome

 

 

Rajeev Chauhan
VCIX-DCV6.5/VSAN/VXRAIL
Please mark help full or correct if my answer is use full for you
Reply
0 Kudos
9990374530
Enthusiast
Enthusiast
Jump to solution

Thanks Rajeev.

I have already tried the sizer tool. Let me gather more information on this.

 

 

Reply
0 Kudos
TheBobkin
VMware Employee
VMware Employee
Jump to solution

@9990374530 

 

Is this an existing vSAN cluster you are existing to expand or is this a cluster currently using internal and/or external SAN resources for storage?

 

Where is your calculation of 120TB required storage coming from and are you including what this data will use if stored as FTT=1,FTM=RAID1 and as thin or thick-provisioned?

 

"in each CPD" Sorry but I am unaware of this acronym, what do you mean by 'CPD' here?

 

1. You should at the very least place hosts in Maintenance Mode if doing hardware changes and validate they see the new storage devices before exiting MM - some storage changes/controllers require cold-reboot and/or other configuration changes to see new devices so likely yes a shutdown may be required.
2. Again, it depends on the controller and device support for hot-swap/hot-plug, not all handle this as gracefully as others and thus planning on possibly needing to shut down hosts to change hardware is prudent.
3. Depends on current storage footprint, how that data is stored, how you plan on storing that data on vSAN (e.g. what Storage Policy rules), your expected data-growth, your expected slack-space (e.g. for temporary snapshots, changes of object structure etc.).

Reply
0 Kudos
9990374530
Enthusiast
Enthusiast
Jump to solution

Hi,

I have checked again and it seems that the hosts do have free drive bays. It's an existing vSAN environment, below is the screenshot where the current capacity of the servers is shown, the approach we are considering is the expansion of 120TB for each of the two Datacenters.

9990374530_0-1679482230155.png

9990374530_0-1679482514118.png

 

 

Reply
0 Kudos