Understand backup to tape

[matteu]

Hello,

I need to use backup to tape for one customer and I have some question to be sure I understand it correctly.

The need :
(1)Backup all the data from all the job (except job dedicated for VM VBM 365 all in one) with this retention : 7d/4w/12m/5y
(2)Backup all the data from VBM 365 job with retention 12m (no daily backup, no weekly, no yearly)

What I create :
for (1) 1 GFS media pool with these settings





for (2) and for the M365 I use these settings (no parallel processing enable to not write on 2 media)





Then I create a backup to tape job

for (1)
I target media pool for (1)







for (2)
the same but target media pool for (2) and GFS scan at 5h and monthly the 1day of the month.


I configure these settings last week and I have different questions :

1) Every incremental use 2 tapes whereas I have enough space on 1 -> I suppose it's because I enable parallel processing and check "enable parallel processing of backup chains within a single tape job"?
Is there a way to use parallel processing only for weekly / monthly / yearly without create other backup to tape job ?

2) The same tape can be used to backup monday and thuesday data. -> Is it because I need to uncjeck this checkbox



3) First weekly happend during last weekend and it used 4 tapes whereas there is space left on first 2. Here I really ignore what happened



Thanks a lot for your answer

[david.domask]

Hi @matteu,

Looking at your media pool settings for the weekly, you don't have append set, so it will always make a new media set when it comes time for a backup to land on the weekly media set. With the parallel settings, the reason it took two tapes indeed is likely due to paralleling; most likely two tasks got assigned and went to a separate drive (the second option in Paralleling allows multiple tasks from the same source job to be parallelized, so that's why it used two tapes likely). After the next run, a new media set was forced.

For your daily, indeed it's likely due to paralleling of tasks again.

Potentially, you can mitigate this by simply disabling task paralleling, but the jobs may take longer and not utilize all the drives as evenly.

[matteu]

Hello ,

I'm not sure to understand. I have only 1 backup job to tape to execute last weekend but I have 2 different media set, I don't understand why. Then, why it used new tape before the actual one has no left free space ? It should backup all the data in one media before take a new one for weekly no ?

Then, I suppose generally you need to take the tape out of the librarie every day. Here, I'm not sure it's done and all the daily backup are on the same drive... Maybe it's better to uncheck "Append backup files to incomplete tapes" to avoid the same tape is used every day.

What I would like is 6 tapes for every day + 2 for the weekly for each week (and then monthly / yearly). With my actual settup, it's not possible because of parallel processing. I will disable it.

I think I will see with my customer how we can do with management of tape.

PS : How can I do when I answer to notify someone with @name in the forum ?

[david.domask]

Sure, I can explain.

Tape paralleling can occur on two levels:

1. Tape jobs
2. Individual backup chains within an added source job that are Per-VM

So consider the following setup:

1 Tape job
Tape Library with 2 drives
3 source jobs added with Per-VM backups (let's say 2 VMs per job)
Job and Backup Chain parallel processing enabled, 2 drives max

Job Parallel processing works on concurrent tape jobs; since there is only one job running, you won't get any benefits from parallel processing there.

But, with Backup Chain paralleling enabled, each Backup Chain from each source job is viable for paralleling.

So when the tape job starts, it will queue like this:

Drive 1: Tape Job a, Source Job A, Backup Chain 1
Drive 2: Tape Job a, Source Job A, Backup Chain 2

If you have two tape jobs, it depends on which locks resources first. Very likely one will always be faster and will lock the media pool.

> What I would like is 6 tapes for every day + 2 for the weekly for each week (and then monthly / yearly). With my actual settup, it's not possible because of parallel processing. I will disable it.

To be clear, you want a specific tape for each day? Or just one tape used regardless?

In general, I think your main issue is just disabling Backup Chain Paralleling: https://zupimages.net/up/22/25/xjfg.png (second checkbox on this image), and you should get the results you want.

[matteu]

Thanks for the explanation


-> disable the second checkbox to allow parallel processing within the same tape job ok I understand it now

I just call my customer now and he want 1 different tape / day.
Then I need to uncheck this to not use the same tape drive for every day right ?



I have an other question about the schedule...

When I use GFS and perform scan at 03am I see the tap job start after the next backup occurs and It's only at 7pm until 11pm. I suppose I can say my gfs scan to start at 6/7 pm right ?
It's like pruning for backup copy if I understand it correctly.
If I want it to work like a "standard" job I need to check like the next picture on advanced properties right ?
What setting do you think is better than the other for my need ?

[david.domask]

> Then I need to uncheck this to not use the same tape drive for every day right ?

Correct, if append is enabled, we will try to reuse the previous medium as long as the media set did not close and the tape has space.

>When I use GFS and perform scan at 03am I see the tap job start after the next backup occurs and It's only at 7pm until 11pm. I suppose I can say my gfs scan to start at 6/7 pm right ?
It's like pruning for backup copy if I understand it correctly.
If I want it to work like a "standard" job I need to check like the next picture on advanced properties right ?
What setting do you think is better than the other for my need ?

Sort of. GFS has a watcher (much like periodic backup copy) and it will start watching at the defined time. It will look for a point created "Today", but if you check that box, it will jump to the most recent point if a "today" point is not available.

[matteu]

Thanks very much for your answer.

I learn a lot on veeam product because of this forum and the veeam community