Active-active or active-standby Media Agent Configuration

Hello @0ber0n 

Thanks for the great question!
I am hoping you could expand a bit further on what you are trying to achieve with the active-passive method.

If it is for a DR/HA logic then you wont gain all that much from having one server passive. If both MA’s are online and both have access to the library then CV will perform load balancing to share the load. If one of the servers go offline then the online Media agent can still do all of the required work without issue.

If you are planning on using Deduplication and having a DDB hosted with one partition on MA1 and another on MA2 then doing active active is also going to be best. A DDB with partitions is not multiple copies of one DDB but a DDB cut into chunks and hosted in each partition. This means MA1 has half of the data and MA2 has the other half. In the event that MA1 goes offline MA2 can still perform all of the required work but you will find that the Dedup rate will go down as all of the signatures in MA1 are missing. 

I feel i went a little to deep to answer your question and i’m sorry if i have just added more confusion so let me just summarize:

working in an active active with both MA’s accessing the same storage and hosting a part of the DDB each will give you the best performance and in the case one MA goes offline your backups and restores can still 100% work. Working in Active Passive costs you performance for the day to day and give you very little gain. It is not recommended. 

Hope this answers your question,

Kind regards

Albert Williams

Hi @0ber0n ,

I agree with @Albert Williams , there is no reason why someone should go with a Active/Passive setup when dealing with MAs.

The CS will balance the load of both MAs and use whatever is available at the time.

Active/Active should be the way to go.

Hi Team,

Thanks for sharing your thoughts.
Yes, what you said is very true, I also asked about active-standby, if active-active cannot be done, can active-standby be possible? After all, I think it is an option too. So, in the light of this information, is there a documentation on how I can set up an active-active Media Agent structure with Commvault?
Can I do this for physical standalone Media Agents, especially those with traditional Windows OS?

Hi @0ber0n ,

You don’t need to do any special configuration for an active-active set up. Just make sure the MAs have a Data Path configured for your Library of choice. CV will automatically perform Load Balancing during backups among all the MAs that have access to the Library.

Hi @Javier,

Is this also valid for SAN attached MA? I know it is valid for CIFS mount path, NAS mount path, Tape Library, but for SAN attached Disk Library? I added the second Media Agent as Data Server-IP over the Storage Pool. But when I want to add a new Data Path via Storage Policy, I get the following error.

Hello @0ber0n 

You cannot add a media agent onto a storage policy that it does not have configured access to the library.
First you must configure the library with mount paths and each of those mount paths have a list of media agents that can access it and how they access it.

The system should automaticity populate the storage policy for you with media agents that have access ( even if you add a new one after the fact to the library ) so you should not have to do any work here.

Please advise if this has answered your questions

Kind regards

Albert Williams

Hi @Albert Williams 

I'm having trouble understanding what you wrote about how to do this configuration. Is there a documentation or steps I can follow to do this? Or is it possible to see how this can be done by having a remote session with an engineer?

Because we really want to do this.

Best Regards.

Hello @0ber0n 

First you would setup your library the way you want it to look with all the Datapath's and access for each of your MA’s: https://documentation.commvault.com/2023e/expert/9298_disk_libraries.html

Then you would setup your storage policy to use that library as its target with the retention rules and such you are looking to use. ( if using deduplication you would set up this to use the library you just built and then point the storage policy at the DDB instead ). 

Now you have created a library with access with both of your MA’s and a storage policy that knows you use both MA’s and will automatically round robin between them. 

I recommend having a go configuration all of this in the Java or Command Center and if you hit any walls or issues a support case may be the best way to get it setup correctly. They will be able to have a remote session with you. 

Kind regards

Albert Williams