Amount of VSAs

Hi @antonio ,

Please check the below documentation, which might be helpful for you. You might need to increase the SCSI controller and VSA here as you are running a concurrent backup job with more than 15 disks.

HotAdd Transport for VMware (commvault.com)

Thanks,

Keshika Bisht

@antonio ,

Thank you for reaching out.

Could you please update for the below questions.

How you are taking backups of the 100VMs,
Individual VMs or Datastore level or Cluster Level.

if you are trying to take 100VMs under single subclient and wanted to take backups are the same time, I would suggest you to increase the number of VSAs. There is no specified number which commvault suggests because it depends on environment to environment. 

For increasing performance you can segregate the VM into multiple subclients and change their schedules.

For data readers options yes you can increase the number of readers too but if you are having only 1 VSA it can effect the performance. 

The Transport mode depends on the configurations and permissions you have given. So to know why the VMs are using different transport mode, kindly check below documentation.

https://documentation.commvault.com/2023e/expert/32037_transport_modes_for_vmware.html

Hello @nishas,

Answering your questions:

• We are taking individual VMs backups with the intention of using subclients to be able to protect these VMs;
• As we are thinking in use more subclients, we also have the intention of use different schedules;
• Implementing more VSAs are also something we are planning, and we want to use LAN Free backups using hotadd transport mode, to minimize the time and use of the resources;

We don't intend to increase the number of data readers to such an exaggerated number, with the idea of using 10 to 15 data readers, and if we notice good performance, we'll keep that number.

Thanks,

Antonio.

@antonio 

Sure, thanks for update.

Taking backup of individual VMs will be helpfully to segragate the subclients.

The intention to increase the Data readers is to find out if single VSA can hold that backups for 100VMs. 

Do update how it goes.

Assuming you are referring to a VMware environment than I would indeed scale-out the amount of access nodes. You can adopt the FREL for this to make life easier and deploy them directly from Command Center. For the actual configuration I would strongly consider adopting protection plans. In case this is AWS than I would look into adopting the scale-out architecture. 

Hi @antonio 

Looks like you are on track and you already received some good tips.

Facing 100+ VM's I would indeed adopt a LAN free method if possible to offload certain aspects.
I would, if HotAdd is the choice, introduce more VSA's, even if it is only to provide a HA implementation of the VSA (Performance wise preferred as well). Then you can put all VSA's in a Client Group and set that group in your Hypervisor Access node list and remove individual nodes, easier to manage 🙂
Ideally one per host so you prevent inter-host traffic, assuming you are using an on-premise hypervisor.

Are you using incremental and synthetic full backups? - can you meet backup SLA’s with incrementals today?

Hi guys, thank all for the good tips and the help so far.

@Onno van den Berg yes, we are in a VMware environment. With your experience, using FREL as access node and VSA, is the performance good? I think using FREL could be a good way out from this customer specially about licensing O.S. in his environment.

@Jos Meijer that’s a good idea! 😀 I think that creating a Client Group can be very helpful to manage and make the things more easier

@Damian Andre Yes, we are using incremental and synthetic full backups. As we are making adjustments in the environment, we haven’t all the workload fully protected yet, the SLA is missing in this moment.