We are currently looking into, redesigning our entire Veeam B&R infrastructure, and the way that jobs are working.
We currently have Veeam Enterprise plus license, for both B&R and Enterprise manager - and we plan to keep both of them.
The environment consists of 5 B&R servers around the world (with alot of proxies), that are all linked to the Enterprise Manager.
Everyday the backup for the given B&R server is performed, and stored on the same site, as the production. After the job is done, the backup is then copied to an alternative location.
This is due to slow uplink at some of the sites, and that we want, to have the backup perform as fast as possible, leaving an minimal impact on the production during work hours.
Considerations:
My idea was to convert everything, to be handled by tags in VMWare, using some standard backup policies. Example:
Daily category:
EMEA_DAILY_7_DAYS
EMEA_DAILY_14_DAYS
EMEA_DAILY_21_DAYS
EMEA_DAILY_31_DAYS
Monthly categori:
EMEA_MONTHLY_3_MONTHS
EMEA_MONTHLY_6_MONTHS
EMEA_MONTHLY_9_MONTHS
EMEA_MONTHLY_12_MONTHS
So to split it up:
EMEA = The location - could be USA, ASIA, EMEA, etc
DAILY / MONTHLY = Indicates how often the backup is performed
Number = Indicates how long retention, will be held for the given job
MONTHS / DAYS = Indicates the value combined with the number above.
Each VM can ONLY be part of one entry, within each category. But ALL VMs will atleast, have 1 DAILY tag enabled.
So VM1 can have EMEA_DAILY_14_DAYS AND EMEA_MONTHLY_3_MONTHS, but cant have multiple daily or monthly tags enabled (This is handled in VMWare).
VM2 can have EMEA_DAILY_7_DAYS, since this is just an frontend server, and can be deployed using an Git repository = the server dont have any "real data" stored there.
The Issue:
Since the daily and monthly jobs are different backup jobs, we dont have an backup chain.
Meaning that when the monthly backup starts, it will pull everything from the production SANs, causing alot of load on the production system.
Ive looked into the Veeam GFS setup, and im a huge fan in general of the concept. But that is done on the job level, for all the VMs within the job.
So if 80% of all VMs in the EMEA_DAILY_14_DAYS job DONT have the EMEA_MONTHLY_3_MONTHS tag, it will still create an backup, containing the 80% of the VMs, that shouldnt be part of this.
From my point of view, there is 2 options.
1.
Create alot of backup jobs, covering our backup strategy:
EMEA_DAILY_7_DAYS_3_MONTHLY
EMEA_DAILY_7_DAYS_6_MONTHLY
EMEA_DAILY_7_DAYS_9_MONTHLY
EMEA_DAILY_7_DAYS_12_MONTHLY
EMEA_DAILY_14_DAYS_3_MONTHLY
EMEA_DAILY_14_DAYS_6_MONTHLY
EMEA_DAILY_14_DAYS_9_MONTHLY
EMEA_DAILY_14_DAYS_12_MONTHLY
2.
From what ive read, Veeam Enterprise Manager can include the Server Version of Veeam Agent.
So we could go with only Daily backup jobs (EMEA_DAILY_7_DAYS,EMEA_DAILY_14_DAYS, etc) within Veeam B&R, and then leave the monthly, quarterly or even yearly backups to the Veeam Agent.
I dont believe we will ever disaster recovery any servers, more than 31 days old. So the monthly backup is only for File-level and maybe MSSQL database-level recovery.
From my point of view, that Veeam agent solution might be the best one, since only 5-10% of our servers, will require more than 31 days of backup.
But im not very familiar with the Veeam agent, and what possibilities there is (can we control the schedule from the B&R servers, can users self-service restore through the Enterprise manager, etc).
We got everything else covered in the upcoming Veeam B&R design, but cant seem to find an "no management approach", to handle our jobs.
So im all ears to any ideas, thoughts, whispers from god, on how to design this!
Sorry for the long post, but would really appreciate some help