Dedup strategy

[jinxnbu]

Hi 

I am setting up a new  BE 2014 envt and using the BE dedup option with HP MSA as disk storage.

Since I am backing up VMs (using API) which are a mix of Linux and Windows does it make sense to have 2 diff disk libraries one as target, 1 for Linux VMs and other for Windows ?

Will this give better dedup ratio than having a single disk library for both VMs ?

Also I have couple of phsyical Wndows servers which have DBs running on it. Any suggestions on what config will give me the best dedup ratio.

3 diff disk library (1 each for Windows VM. Linux VM, phyiscal servers )?? or have all of them point to a single disk library ?

Thanks in advance. 

[CraigV · Accepted solution]

Hi,

You can only have 1 dedupe target per media server. Just split the backups off from each other (LInux/Windows).

Read up on the BE 2014/15 dedupe best practices here:

https://support.symantec.com/en_US/article.howto74446.html

THanks!

[pkh]

It does not matter from which server the data comes from. If a data chunk is found in the dedup folder, it would not be stored. See my article below. https://www-secure.symantec.com/connect/articles/deduplication-simplified-part-1-backup

[jinxnbu]

Thanks pkh and CraigV for your responses.

@CraigV .. you mentioned split the backups off from each other. If i understand correctly, I should/can create 1 dedup target disk library and have 1 vmware job to backup only windows VMs and another vmware job to backup only linux VMs. ?

I am reading a lot of articles on forums where ppl are encountering issues with GRT enabled app backups using agent for vmware to dedup storage.. any insight into if this is the case for BE 2014 as well or is it better to proceed with remote agent backups (native approach) to dedup storage ?

Just to reiterate the queries I have

1) For dedup target is it better to go with agent based backups or Vmware backups (including GRT for DB)

2) If running backups using agent for vmware, i intend to use the san mode of zoning the datastore to the media server after running automount disable and scrub. I have used this method for non dedup disk storage however for dedup storage is there any additional points to take care of or would it still be the same approach. 

Thanks in advance

[Gurvinder · Accepted solution]

1)  Keep Linux and Windows VM Backup in seperate Jobs since Linux we do not support GRT backup with Agent for Vmware Virtual Infrastructure

Eg :

File Server Backups  - Use AVVI Backup with File Folder GRT enabled

Application Server like exchange, SQL., AD - Use AVVI with File Folder and Application GRT

Linux VMs - Use Agent Backup or AVVI Backup whichever suits your Backup and Restore strategy. With Agent you can perform individual file restores easily else with AVVI Backup of Linux you will have to restore the entire VM to get that 1 file. So depends which one you choose for Linux VM backup.

2) The way you perform AVVI Backups to normal storage is the same as you would do AVVI or any other backups to Deduplication Storage.

Note : Backup Exec has stream handler which easily understand which part of vmdk has changed and hence provides better dedupe ratio. This works when GRT is enabled for the Backup of VM using AVVI.