Is there a Tape Backup white-paper anywhere?

[pkelly_sts]

I'm curious if there are any such white papers/best practice documents out there as I don't think the product documentation covers anywhere near enough real-world scenarios to be of much use to a lot of people.

I've been hacking at (testing) tape backups for months now & still don't feel I understand what should/will be written to what tape/media set in a given scenario and I really don't think my needs are that complicated.

I've been dealing with Backup Exec backups through DAT, Exabyte 8mm, DLT, AIT & finally LTO for around 16 years now and Veeams tape "logic" still totally eludes me months after first looking at it!

I'm sure that once I understand it I'll smack myself on the forehead as I'm sure it really /can't/ be that difficult & I'm desperate to switch away from Backup Exec but have absolutely NO confidence whatsoever that I can say, hand-on-heart, that on THIS day, THESE files will go to THAT tape etc.

Would love to be pointed towards any available resources (except the online documentation which really isn't sufficient in my opinion).

[Shestakov]

Hello Paul,
We have Veeam Backup & Replication Best Practices book where is a chapter about Tape Support. Hope that`s what you are looking for.

[Dima P.]

Hi Paul,

Have you seen the Overview part of the Help Center? It’s more about engine and describes the internal logic of the every single feature inside Veeam B&R. I am asking is because people usually go to ‘Administration’ part skipping this one.

For instance, GFS tape job logic article looks very sharp.

[pkelly_sts]

Hi Dima, I believe I have gone through the overview section but I shall do so again, having gained some personal insight/experience with it (it often never helps reading the "manual" before you've used the product much!).

However I have an excellent case in point right now having agonised yesterday over how I would like to manage tape backups.

To support my testing (of both SOBR and tape) I'm running a second, completely independent, parallel backup chain which is a Forever Forward job, currently configured with 14 RPs but which, as of yesterday, had only been running for a week so was up to 7 RPs as of yesterday afternoon.

After much deliberation, I decided the following:

1) That I would like to run a daily tape job, doing either a synthetic or active full (not decided yet) at the end of each week, probably on Friday or Saturday (Note: I already have off-site copies for now)
2) I would like subsequent incrementals to go to the same physical tape so each week a tape should consists of the full .vbk then generally 6 x .vib
3) I will start a new media set each week on the same day as I run the full backup, this should ensure that the chain is kept together
4) I would like to rotate these through 8 tapes, so therefore I should be able to maintain 56 RPs on tape quite easily
5) To achieve this, I'll configure an overwrite protection of 8 weeks (for now, this can get slightly more complicated if jobs run slightly late etc. but I'll re-visit that later)

So, looking at my (per-VM) test backup set, I currently have one .VBK and 7 .VIBs per VM, representing approximately 7 days of backups (I might have paused it for a day so the range might span a little more)

Knowing that my full backup as a single .vbk comes in around 1.6Tb or so, and my incrementals average about 25Gb/day, even taking into account 10% larger due to per-vm, I should easily fit this set onto a single tape as first run job.

I configure a media pool as follows:

1) To process BOTH full AND incremental backups
2) To create a new Media Set each Saturday

I configure the job to run yesterday evening, expecting the following:

1) Create a new media set as this will be the first media set in this pool
2) To copy the .vbk files to the tape (1 per-vm, somewhere around 1.6-1.8Tb)
3) To copy the .vib files to the tape (7 per-vm, somewhere around 175 - 250Gb) So *total* on tape should perhaps be slightly over 2Tb worst-case
4) To have enough space on the tape to sqeeze on another few incrementals between now & Saturday as an exception
5) To ensure I have another blank tape available for Saturday when a synthetic full (to tape) is configured

What I actually found when I came in this morning was that the job had so far processed, read & transferred 2.4Tb, waiting for a second tape, and was *only 70% complete* !!!

I then immediately reviewed both simple repositories that make up this SOBR to check their sizes on disk: Repo1 has 635Gb, Repo2 has 1116Gb - TOTAL 1751Gb !!!

So how on earth can I be up to 2.4Gb already at only 70% completed???? IT MAKES NO SENSE!!!

When things like this are happening, I can make the most detailed plans in the world but they mean absolutely nothing in the real world so I really do not know what to do ((

[Edit: Just to confirm, I've just gone through the stats for the original source job and confirm the following sizes written to disk, as reported by the job itself:]

Full 1.6Tb
Incr 1 26.5Gb
Incr 2 5.9Gb
Incr 3 23.7Gb
Incr 4 21.3Gb
Incr 5 23.9Gb
Incr 6 22.8Gb
Incr 7 36.1Gb
---------------------
Total 1789Gb

[pkelly_sts]

Looking at the size of the data being copied to tape, it looks suspiciously to me like it's writing a full backup twice so I had a trawl through the log file and found the following lines in various places:

[01.06.2016 01:47:38] <01> Info Last Synthetic Required Only = True

[01.06.2016 01:47:49] <01> Info [CTapeSourceBackupJob] Backing up vbk again because it's a first time backup

[01.06.2016 01:47:45] <01> Info [CTapeDailySynthesizedBackupCalendarSpecifics] Only latest virtual full backup is required. Skipping older virtual full backup creation

Why on earth, on a first-pass, would it be creating a synthetic full, on top of the full that is already there, ready to write to tape???

[pkelly_sts]

I also just came across this post from you Dima. Is it still relevant?

tape-f29/backup-to-tape-performing-unne ... ml#p149616

[Blue407]

you're not alone!

I too am coming from BE, a product that I found very logical. I find the tape backup side of Veem odd. I also feel I'm missing something obvious. It feels like a product bolt-on that's only designed to work a certain way.
I'm used to being able to easily append to tapes, that doesn't seem so easy in Veeam.

[pkelly_sts]

Glad it's not just me :-/

It seems to me that appending is easy, if you want to append forever to a single media set, but it doesn't make sense to have a "media set" that spans months etc. but sometimes a weekly media set is too often as well. There just doesn't seem to be nearly enough flexibility.

[Blue407]

I want to be able to simply run a 2nd job the same day to append to the tape that the first job used. Sounds simple...
So firstly backup critical servers, then non-critical.
Thats one thing BE did well, have a format option for the first job, then the rest just append.

[pkelly_sts]

Logic would suggest that you simply point the job to the same media pool, and the media pool should be configured with media set settings of "Do not create, always continue using current media set" but that would append "forever".

If you wanted to force it to start a new media set with some kind of regularity then you'd perhaps set the 3rd option: "Daily, on these days"

Then again, I could be totally wrong based on my own experiences! LOL!