Skip to main content

I’ve tried reading the docs and its deep, and I’m not really finding anything usiful…

Background:

We have several Vmware backups occurring of VM’s, and there is this “Deduplication” setting on the (client) → Virtual Server → VMware → (backupset) → subclient (right click) → Properties: htab] Storage Device isub tab] Deduplication where you can chose “On Client” or “On MediaAgent”

We are using proxies for all VM backups (if that matters).

For *most* of our VM backups, the “Deduplication” setting mentioned above is set to “On MediaAgent”, but a few are set to “On Client”. here appears to be no rhyme or reason these are set one way or another. For all of these, these are vCenter/VMware hosts that are physical and we own (local on the internal network).  I could see that “On MediaAgent” would be an OK setting, as everything is local and the MediaAgents are very powerful compared to the proxies.

I have just set up a new VMware backup, for a vCenter/VMware instance that is in the cloud. The default (on config/install of the “VMware Vcenter Client”) for the “Deduplication” settings is “On Client”. 

Questions:

  • Since this is a “cloud” VMware/ESX system, with the proxies (as VM’s) performing the backups in the cloud… is Deduplication “On Client” a better setting than “On MediaAgent”?  I’m thinking of not only “proxy CPU power” but also “data sent back to the CommVault system” from the proxy in the cloud.
  • Are there any gotcha reasons for using “On MediaAgent” vs “On Client.”?  I would figure that since ‘On Client” is the default, this is the better default option for most cases. I’m not sure if this causes more disk space to be used up on the client or if there is a “gotcha” (meaning: something very obvious you need to take into consideration with this setting, like massive CPU usage, storage usage, memory, etc.)

Hello @tigger2 ,

 

Thank you for reaching out with your qeustion.

To address your questions in turn:

  1. The use of On Client vs On MediaAgent Deduplication is dependent less upon the client build type and more on your infrastructure build.  The setting for “On Client” set at the subclient level allows for the generation of signatures on the client machine, versus on the MediaAgent.  What this means is that instead of sending every piece of data over the network to the MediaAgent to Deduplicate, the client will instead generate the signature locally, then send it to the MediaAgent to compare against the Deduplication engine.  If the MediaAgent recognizes the signature, further data is not transmitted for that block.  If the signature is new, it asks the client to then send the entire block.  If the setting is set to On MediaAgent, the client transmits the entire data block to the MediaAgent every time.
  2. Regarding “gotcha” concerns.  If the “On Client” setting is enabled only, with no source side caching, then there is no significant additional strain on the client regarding resources.  However if this setting is enabled in tandem with Source Side Caching, then each time a signature is generated on the client, it also stores a copy of that signature locally.  While each signature is not very large, they can eat up space on the cache disk if the signature list grows too large.

For more details please reference the following documentation:

https://documentation.commvault.com/2022e/expert/58533_storage_device.html

https://documentation.commvault.com/2022e/expert/12451_modifying_deduplication_properties_for_client.html

https://documentation.commvault.com/2022e/expert/12401_optimize_storage_space_using_deduplication.html

Let me know if you have any other questions.

 

Regards,

Josh P

Hi, Any limitation on number of VMs per subclient?

Hi, Any limitation on number of VMs per subclient?


I’m not sure if there’s a hard maximum.  You will probably have more data than your Storage Pool is designed to manage before you hit a # of VMs issue.

Thanks,
Scott
 

Hello @Scott Moseman and @Josh Perkoff 

Im also write in this thread, because Im wondering about using Client-Side Deduplication without any Commvault VSA.

Networker for example works only with Client-Side Deduplication, so I’m try to find Commvault’s  best practices about Client-Side Deduplication.

Thank you in advance,
Nikos

Im also write in this thread, because Im wondering about using Client-Side Deduplication without any Commvault VSA.


I do not understand the question.  You will need VSA to backup VMs.

Thanks,
Scott
 

Im also write in this thread, because Im wondering about using Client-Side Deduplication without any Commvault VSA.


I do not understand the question.  You will need VSA to backup VMs.

Thanks,
Scott
 

Hey @Scott Moseman 

Thanks for your reply!

Im actually looking for pros and cons using Client-Side Deduplication, instead of MediaAgent Deduplication in a Vmware environment.

The recommended / default method for Commvault I assume is MediaAgent Deduplication , right?

Thank you,
Nikos

The DDB is hosted on your MAs, so you’re deduping on the MAs either way.  If you have VSA installed on your MAs, the concept of client side dedupe doesn’t really apply.

If you have standalone VSA proxies, enabling client-side dedupe means potentially less data being transferred from the hypervisor to the MAs.  The VSA proxies would incur some additional load and space consumption (dedupe cache), but save you some bandwidth.  Which route you go probably depends on which resource is more constrained.

Thanks,
Scott
 

Reply


Terms & ConditionsCookie settings