Backup concept

[redhorse]

Hello,

we still use Backup Exec to copy our Veeam-Backups to tape but now it is time to change this and move everything Veeam. So forget about our current backup concept, this is what we want to do in future:

- Daily backup to fast disks with short retention (7 restore points for 7 days)
- Backup copy to slow disk with long retention (60 restore points for 60 days)
- Daily incremental backup to tape with weekly full with
-- 5 weekly backups on tape
-- 12 monthly backups on tape

How can we achieve this?
I suggest we do a daily forever incremental backup with retention period for 7 days for the backup to our fast disks. Then we need a daily Backup copy job with retention for 60 days to our secondary repository, is this possible? Is there a need of a full backup or does it also work with forever incremental?
How does it work with the tape backup?

[Shestakov]

Hello,
Your reasoning is correct.
You need a basic backup job + backup copy job(also consider GFS retention) and backup-to-tape job.

Then we need a daily Backup copy job with retention for 60 days to our secondary repository, is this possible?

Yes, it is. You can either use simple retention with 60 points or GFS. Backup copy jobs use forever incremental method by default.

How does it work with the tape backup?

You just need to create backup-to-tape job to copy backup files to the tapes. Are fast disk, slow disk and tape media placed on the same site?
Thanks!

[redhorse]

Thanks for your help & fast reply!

Yes, it is. Backup copy jobs use forever incremental method by default.

Could it be a problem with backup consistency when we have 60 restore points with forever incremental? Is it necessary to use Sure Backup to check the backups on both repositories?

You just need to create backup-to-tape job to copy backup files to tape. Are fast disk, slow disk and tape media placed on the same site?

Yes, they are on the same site. So I copy the backup files from the fast repository to tape, that right?

[Shestakov]

Could it be a problem with backup consistency when we have 60 restore points with forever incremental? Is it necessary to use Sure Backup to check the backups on both repositories?

60 point-long chain should not cause a consistency problem. Surebackup doesn`t imply working with backup copy job files. There is an integrity check, so, it should work fine.

So I copy the backup files from the fast repository to tape, that right?

That`s correct. But the best practice is to have data copies in at least 2 sites in a case of disaster. I would consider backup copy to another site.
Thanks!

[redhorse]

60 point-long chain should not cause a consistency problem. Surebackup doesn`t imply working with backup copy job files. There is an integrity check, so, it should work fine.

Okay, that sounds good. So never have to do any checks with backup copys to be sure that they are working as long as there is no issue with the integrity Check?

That`s correct. But the best practice is to have data copies in at least 2 sites in a case of disaster. I would consider backup copy to another site.

Yes that's my oppinion, too. This will be the second step.

Just one more point: We still have one physical server which we want to backup with Veeam Endpoint to our repository with backup copy and backup to tape, I think that's just an easy thing. But how can we check the integrity of our Veeam Endpoint Backup? We want to use Sure Backup to check the backups of VMs but this is not possible with Endpoint backups, isn't it?

[Shestakov]

Okay, that sounds good. So never have to do any checks with backup copys to be sure that they are working as long as there is no issue with the integrity Check?

Correct.

But how can we check the integrity of our Veeam Endpoint Backup? We want to use Sure Backup to check the backups of VMs but this is not possible with Endpoint backups, isn't it?

Not in v1, since there is no P2V restore to the infrastructure.
Thanks!

[redhorse]

Okay, now it is clear for me what to do. What chances do I have to check if the endpoint backup is valid?

What is the best way to configure my scenario? I guess it should be something like starting the sure backup job after a successful backup on the fast repository and if sure backup is successful starting the backup copy and backup to tape job. How often do I have to run the sure backup job?

[Shestakov]

As far as I remember we had a similar conversation 6 months ago.
Nothing has changed. Ideally you make recovery verification for each reatore point before backup copy job run. So your plan is good.
Thanks.

[redhorse]

As far as I remember we had a similar conversation 6 months ago.
Nothing has changed. Ideally you make recovery verification for each reatore point before backup copy job run. So your plan is good.

Thanks, I forgot that we already discussed that point. We have about 35 VMs and I am not sure if we have enough time to check every VM and every restore point each day. There was a recommendation how often an active full backup is necessary when using reverse or forward incremental, so I thought there is also a recommendation how often a sure backup should be done to ensure that the backup is healthy. It's clear that a daily sure backup is the safest way but it is not realistic in each environment.

By the way my plan is to use our ESX-SAN (EqualLogic) with an own LUN via iSCSI as the first backup repository and a slower NAS-Storage (Synology) via SMB for the backup copy job. The first repository will be copied on tape once a week as a full backup. Is that okay from your point of view?

Our EqualLogic is attached to our backup server via iSCSI and we use direct SAN access so there should be a fast communication to get the files from ESX and fast communication to write it back to the backup LUN. The backup copy NAS is reachable via Gigabit LAN.

[Shestakov]

Your infrastructure and plan look good. The only recommendation, as I earlier said, is to have one copy of backup files offsite.

[redhorse]

I did some tests now and everything seems to work fine. I created a SureBackup job which starts automatically after the backup job, is there any way to start the backup copy job automatically when SureBackup is finished without error?

[Shestakov]

There is no way to link it in UI. However you may try to use a custom script.
Thanks!

[redhorse]

Okay. Is it possible to do SureBackup and backup copy at the same time with the same VMs?

Thanks!

[Shestakov]

Yes it is, since both jobs work on read.

[redhorse]

Yes it is, since both jobs work on read.

Thank you very much for your patience, things are now clear for me!

[Shestakov]

You are welcome!
Once you have a problem, don`t hesitate to ask!

[skrause]

Our EqualLogic is attached to our backup server via iSCSI and we use direct SAN access so there should be a fast communication to get the files from ESX and fast communication to write it back to the backup LUN. The backup copy NAS is reachable via Gigabit LAN.

Is that EqualLogic running your production VMs as well?

That is the only part of your configuration that throws up a red flag to me. If you have an issue with your production SAN, your most recent backups will also be effected.

Also, if you can hold off until v9 for your tape stuff you will be MUCH happier. v8 tape works, but it is somewhat kludgy to do what you want (we do something similar only on a monthly basis). They are pretty much totally revamping the tape features in v9 and it will be a ton better.

[Shestakov]

skrause is correct.
You can read more about new tape features in v9 here.

[redhorse]

Is that EqualLogic running your production VMs as well?

That is the only part of your configuration that throws up a red flag to me. If you have an issue with your production SAN, your most recent backups will also be effected.

Yes it is, but the repository for backup copy is connected via Gigabit and in case of storage issue I am able to restore from a repository located on the same site. So I thought it is not much better to have a separate repository for the main backup and using the production SAN is much faster than a separate cheaper one. Or am I wrong?

Also, if you can hold off until v9 for your tape stuff you will be MUCH happier. v8 tape works, but it is somewhat kludgy to do what you want (we do something similar only on a monthly basis). They are pretty much totally revamping the tape features in v9 and it will be a ton better.

Thanks for your advice. Is it difficult to configure my setup in v8? I read something about a better GFS tape support in v9.

[Shestakov]

Thanks for your advice. Is it difficult to configure my setup in v8? I read something about a better GFS tape support in v9.

GFS logic can also be implemented on tapes in v8 with the following workaround:

Create a media pool called “Weekly” with retention period equal to X weeks.
Create a backup to tape job called “Weekly” and point to the “Weekly” media pool and schedule it to run on weekly basis.
Create a media pool called “Monthly” with retention period equal to Y months.
Create a backup to tape job called “Monthly” and point to the “Monthly” media pool and schedule it to run on monthly basis.

Same for "Quarterly" and "Yearly" retention.
More information can be found here.

[skrause]

Yes it is, but the repository for backup copy is connected via Gigabit and in case of storage issue I am able to restore from a repository located on the same site. So I thought it is not much better to have a separate repository for the main backup and using the production SAN is much faster than a separate cheaper one. Or am I wrong?

You are not necessarily wrong that your production SAN will be faster than a separate cheaper one. The issue is, as one Veeam SE said at the Users Group meeting at VeeamON "If you are storing your backups on the same storage as production, it isn't really a backup." If your production SAN fails, you lose not only your production data but you lose the ability to easily restore from backups so best practice is always to use a different storage device for your backups. And it doesn't need to be a SAN, a fairly basic server with a good RAID controller and a lot of high capacity disks is a good way to do things.

[redhorse]

I just want to give a feedback to you:

I decided to buy a new Server (Dell PowerEdge R330) with enough space for the primary backup (5 restore points), this is a dedicated server for Veeam (Version 9) with all Veeam services. We also backup our physical domain controller and the backup server itself via Veeam Endpoint Protection to that repository. In addition to that we use a NAS (Synology) as a backup copy target for all backups with CIFS (60 restore points). We use forever forward incremental with daily Sure Backup for verification.

Now it is time to move the tape backup from Backup Excec to Veeam and I am still not sure how to implement GFS because I can’t use the new GFS Feature because I have a File Copy Job which should also be stored on the GFS tapes. As I understood this is not possible because GFS is only supported when backing up VMs, is that correct?

Shestakovs concept is clear for me but I think there will be two tape jobs at the same time when weekly backup meets monthly backup and I can’t avoid it, that’s also correct?

[Shestakov]

Hello,
Thanks for writing back.

Shestakovs concept is clear for me but I think there will be two tape jobs at the same time when weekly backup meets monthly backup and I can’t avoid it, that’s also correct?

That`s correct. The way to avoid it is to leverage the new tape GFS feature, where you will have a single write on a tape, which saves both time and space on the media pool.
Why do you use file copy job instead of backup ones?

[redhorse]

That`s correct. The way to avoid it is to leverage the new tape GFS feature, where you will have a single write on a tape, which saves both time and space on the media pool.
Why do you use file copy job instead of backup ones?

Sorry, I mean "File Backup to Tape", the files are stored on a NAS. But that's the same problem, isn't it?

[Shestakov]

Correct, File to tape backup doesn`t work with GFS media pool due to different synthetic full logic.
So why not to use backup to tape? Since you leverage Surebackup, I can guess the backup to tape option is also available for your product edition.

[redhorse]

Correct, File to tape backup doesn`t work with GFS media pool due to different synthetic full logic.
So why not to use backup to tape? Since you leverage Surebackup, I can guess the backup to tape option is also available for your product edition.

Yes we have Veeam Enterprise and I want to store all VMs + some files from a share (no VM) to the GFS-Tape, so perhaps I understand you wrong because I don't know how to use backup to tape for it?

[Shestakov]

Backup to tape is convenient for VM backups. Especially with new GFS Media pools. You can use file to tape with a simple media pool as well.

[redhorse]

Backup to tape is convenient for VM backups. Especially with new GFS Media pools. You can use file to tape with a simple media pool as well.

Okay but then I need a separate tape for the file to tape backup which is about 50 GB that's not what i want. Is it possible to use simple media pool and move the last synthetic full manualy to a separate media pool which has one year retention? What would happen with the following incremental backups?

[Shestakov]

You can write fulls and incrementals to different simple media pools and set a 1 year retention on the media pool with full backups. You can also write file to tape job on the tapes with increments.

[redhorse]

You can write fulls and incrementals to different simple media pools and set a 1 year retention on the media pool with full backups. You can also write file to tape job on the tapes with increments.

What I mean is this:

1. Create a Media Pool called daily, synt full weekly and Daily increment, weekly new media set, retention two weeks
2. Create a Media Pool called monthly, no jobs, retention one year
3. Move the media set created on last saturday in month (just the synt Full Backup) to the monthly pool (not automatically)

Is that possible? What happens with the retention after moving the media set? What happens with the next increment when the full backup is not in the original media pool anymore?