This post is more than 5 years old
51 Posts
0
5621
Using DD as target for checkpoint backups - size issues
Hi,
have any of you tried to use DD as target for checkpoint backups on single node Avamar installation?
I tried this out of curiosity with Gen4 3.9TB node with Avamar 7.0.1-61 (upgrade path 6.0.1-66 -> 6.1.1-87 -> 7.0.1-56 -> 7.0.1-61) and DD2500 with DDOS 5.4.1.2.
Avamar node was 84% full and DD2500 was almost empty (80gigs of other data) - I was surprised by the size of the checkpoint backups as they took 3.2TB from DD.
Those two peaks are successful checkpoint backups and then I disabled it when I discovered the huge size it was taking. Notice that it took 9 days to get them out from DD, don't know if there is some command that could do this manually?
Have others received this kind sizes? I'm wondering could the issue be that the this node is not natively installed with 7.0 as it was installed with 6.0 and then upgraded to 7.0.1.
ionthegeek
2K Posts
1
March 19th, 2014 07:00
My pleasure!
Checkpoint backups are designed for use cases where it's not practical or economical to set up a replication partner. Replication is generally the better option.
The MCS database is backed up to the /MC_BACKUPS account once an hour. This account is replicated by default and can be reverse replicated and restored from the replication target if the source system is lost. You should never have to back up the MCS database yourself.
ionthegeek
2K Posts
1
March 19th, 2014 06:00
This is expected. Checkpoint backups have to back up all the data containers on the Avamar. The total size of all the containers on most systems will be at or near the licensed capacity of the system (3.9TB in this case). Checkpoint backups also de-duplicate relatively poorly since the data inside the containers is already de-duplicated and the containers change frequently.
niko.virta
51 Posts
0
March 19th, 2014 07:00
Thanks Ian,
size of the checkpoints just surprised me ... I was thinking of getting few 7.9TB nodes among this one to get backed up but as it would take total 10-15 TB from DD and it is bit too much.
If I need data redundancy I will replicate that data to other node/DD. If node dies I can't afford losing MCS data. Been backuping MCS database with scripts and seems like I have to continue that way.
niko.virta
51 Posts
0
March 19th, 2014 08:00
Yeah, I'm backing up MCS database only if the node is not replicated - it goes nicely with same script that collects all logs from Avamar to centralized logserver for longer retention.
ionthegeek
2K Posts
0
March 19th, 2014 10:00
Technically speaking, it's mandatory to have either replication or checkpoint backups on production systems...
If you wanted to set up a supported solution for preserving the MCS backups, you could stand up an AVE to serve as a replication target for MC_BACKUPS, EM_BACKUPS and AVI_BACKUPS for these unprotected systems. My concern is that if you're using a custom script to save the MCS database, you may run into issues if you ever need to restore it and support would not be able to help.
ionthegeek
2K Posts
0
March 19th, 2014 11:00
Oh, right. I always forget about those.
Just having the MCS database isn't sufficient, unfortunately. You'll keep the datasets, schedules, etc. but the client accounts in the mcdb are tied to corresponding GSAN accounts by their client ID and if you lose the node, you lose the GSAN user accounting system. Replication preserves this GSAN account information. We can talk about the details in another thread if you'd like.
niko.virta
51 Posts
0
March 19th, 2014 11:00
Ian,
Not replicated nodes are Business Edition nodes.
http://emcinformation.com/130402/LP/82cfca21-6495-44fe-a294-975d4e572aab.ashx?topic=Feature
"It provides a conveniently-sized (7.8TB), turnkey, affordable, deduplicated backup system with optional (not required) replication."
If you have best practices for backing these up, more than taking that MCS database dump outside Avamar node, please share. Maybe in new thread?
niko.virta
51 Posts
0
March 19th, 2014 13:00
Thanks Ian, opened up another thread for BE metadata backups.
https://community.emc.com/thread/191078
ldezorzi
2 Posts
0
January 24th, 2016 02:00
Hi Niko,
I have the same problem as you.
Following a crash of one of my 2 avamar nodes, EMC rebuilt the node but left the backup checkpoint on the DD.
How do you disable target avamar checkpoint backup ?
Thanks for you reply.
ionthegeek
2K Posts
0
January 25th, 2016 06:00
Enabling or disabling checkpoint backup is covered in the Administration Guide.
The instructions are on page 356 of the Avamar 7.1 Administration Guide:
https://support.emc.com/docu53936_Avamar-7.1-Administration-Guide.pdf?language=en_US
Finding the page number for other versions is left as an exercise to the reader.
ldezorzi
2 Posts
0
January 25th, 2016 06:00
Thanks Ian.
I did not remember that we could set parameters of DD from Avamar. I looked for from Avamar.