dedup and replication question for 2 vaults
Hi
I have 2 vaults one on site and one off site. Both are managed and right now both have dedup enabled. Dedup is set to "at target" for all my plans in order to load the backup machine rather than the source machines with deduplication. As i'm observing the backup tasks, i've noticed that the 1st vault (on site) pulls all the data from the source machines then dedups it and then sends it to the 2dn vault (off site). However when it sends the data to the 2nd location it appears to be sending it "raw", so at the very end of the job the 2nd vault attempts to deduplicate as well.
My original understanding of the process was that the deduplication would only happen once at the on site vault and only the reduced amount of data will then be replicated over to the 2nd vault.
So, there're 2 problems i may be having, one is that my second vault is attached to a rather slow old machine with 4gb of RAM and it if it has to deduplicate the data it'll hang up all my plan schedules, and the other one is, if the data really travels raw from individual machine images rather than from an already deduplicated set, that would create excessive traffic for me.
From inner workings standpoint what are the consequences of me enabling or disabling depup on the 2nd (remote) vault?
thank you
R U kidding? Acronis un-deduplicates the data and sends across the WAN fully hydrated?
The whole point of deduplication is to reduce WAN bandwidth usage, speed up replication, and of course, decrease disk consumption.
Please tell me you mis-spoke and that Acronis really does replicate only the changed (deduplicated) bytes!
I don't want to have to migrate to yet another backup solution. :(
Thank you!
Hello John, Anna was correct, however the process is more difficult. Replication from 1 dedup vault to another dedup vault is not going to transfer unified_data file as if it was rsync. However the if you have deduplication on source enabled then we won't be transferring all the data that you have.
Let me give you examples.
Let's say you have onsite Storage Node with dedup vault: ASN1 and you have offsite Storage Node with dedup vault: ASN2.
You have agent in the same network as onsite ASN1. You backup that agent to ASN1.
Your scenario: Since you are backing up and replicating in 1 plan the Option to disable "Deduplication on Source"is going to apply for Replication process as well.
1. So your scenario (with dedup on source disabled) will do this:
Agent will backup to ASN1. Since dedup on source is disabled the agent will send all data to ASN1.
ASN1 then deduplicates the data on itself. Once the backup finishes replication starts. Replication undeduplicates the data and makes the instance of undeduplicated backup and starts binary replication of the data from the undeduplicated backup to the ASN2. Since dedup on source is disabled then ASN1 will send all data to ASN2. ASN2 will deduplicate the data on itself.
2. If you have dedup on source enabled.
Agent will backup to ASN1. Since dedup on source is enabled the agent will send only unique data to ASN1. Once the backup finishes replication starts. Replication undeduplicates the data and makes the instance of undeduplicated backup and starts binary replication of the data from the undeduplicated backup to the ASN2. Since dedup on source is enabled then ASN1 will send only unique data to ASN2.
In this case the network load is decreased.
ASN will undeduplicate the backup before replication in any case because we support replication from dedup vault to non-dedup locations. That process is local, however it takes a lot of resources. So that will require even more resources than we have specified in our Best Practices:
That is why replication from deduplicated vaults is not a very recommended feature.
Thank you for the detailed explanation!
Fortunately we do utilize source deduplication, so we are following best practices.
Our replication slowness must be caused by something else.
We are working a support call on this now.
Thanks!