Synthetic Full Backup

[leockw]

Dear all,

I'm new to V&R and I would like to clarify some concept on Synthetic Full Backup.

1. As I know, in a job setting synthetic full backup only can select the day in a week, can I use Synthetic Full Backup for monthly backup(eg.1 full + 29/30 incremental)?

2. If no synthetic full backup for monthly, then, when I make a Backup Copy Job as offsite backup, the backup copy job still need to copy all weekly full backup to offsite?

3. If use backup copy job for a backup job(full->incre) to offsite, the first full backup need copy to offsite over one day and overlap the incremental backup job, will it waiting the "full backup" to offsite completed and start to copy the "incremental backup" to offsite?

Leo

[PetrM]

Hi Leo,

1. Basically, you may configure GFS retention policy, monthly backups will be synthesized from already existing backups.

2. Backup copy does not copy backup files but always transfers the actual state of machines in backup. This actual state consists of blocks from the latest incremental chain, for example: last full + incrementals.

3. If incremental backup job run is started and locks a restore point on source while backup copy is running: backup copy job pauses data transfer, waits for incremental run completion and resumes transfer.

Thanks!

[foggy]

Hi Leo, welcome and please review the answers below.

1. Monthly synthetic fulls are not an option, here's another existing thread on that matter.
2. Backup copy job doesn't copy entire backup files but rather only blocks required to build the new restore point on a target, regardless of what the latest source restore point on source is - full or incremental.
3. Yes, the backup copy job will catch up in case the initial full synchronization takes longer than a backup period.

[leockw]

Dear PetrM and foggy,

Thanks a lot for your explanation.

Is that means backup copy job is use to backup the local backup state to offsite?

Can I can setup the job as following,
Local backup:
Synthetic weekly(full + incre) + Monthly(weekly full)

Offsite backup(backup copy job the local backup state?):
Monthly(full + incre) + Yearly(monthly full)

Leo

[foggy]

Yes, the backup copy job uses backups created by the regular backup job as a source.

Not sure I fully understand the setup you suggested, but backup and backup copy job chains are totally separate and you can configure them independently.

[leockw]

Dear foggy,

I just wonder if I use synthetic weekly full backup and keep 9 week on local backup without monthly and yearly as following,

Sun Mon Tue Wed Thu Fri Sat Sunday
Incre Incre Incre Incre Incre Incre Full Incre

Then, I setup backup copy job but without weekly backup, only keep 6 monthly and 7 yearly, what will I have in the target backup repository?

As I have setup a test backup copy job on last Wednesday and I have the first full backup and 5 incremental now, is it will "generate" a monthly full backup on the coming June?

Leo

[foggy]

Backup copy retention is independent of the source backup job. Once it reaches the configured simple retention (the number of restore points in the regular chain, i.e. full + increments), it will start generating GFS restore points (monthly and yearly in your case) according to the specified setting.

[leockw]

Dear foggy,

That means the backup copy job will do the full backup copy job(not synthetic/combine?) on the source backup repository through the network on the coming June?

If I keep the weekly full for the backup copy job, it still do the full backup copy job(not synthetic/combine?) on every Saturday through the network, am I right?

If this is correct, seems the network usage will very high.

Leo

[foggy]

As mentioned earlier, the backup copy job always copies only changed blocks, regardless of what type the source restore point is - full or incremental. GFS restore points do not require copying the entire full unless you explicitly select the 'Read from source...' check box.

[leockw]

Dear foggy,

I have a little question on backup copy job again,

1. I found the backup file health check consume a lot of bandwidth, is it really need to check?

2. I would like to know what is the usually bottleneck on backup copy job for offsite? It is because I found 1.7T GFS synthetic full using over 12 hours, is it caused by CPU power, memory or bandwidth?

Kind Regards,
Leo

[Mildur]

1. I found the backup file health check consume a lot of bandwidth, is it really need to check?

It depends. Do you want to have reliable restore Points when you need it? Health check is for finding corrupted data blocks in you restore Point. If you need to restore data, you want to be sure that the backups are not corrupted. You can disable the health check if you are doing regular SureBackup Jobs or manually test restores. I don‘t recommend it to disable it, if you don‘t have any alternative tests.

2. I would like to know what is the usually bottleneck on backup copy job for offsite? It is because I found 1.7T GFS synthetic full using over 12 hours, is it caused by CPU power, memory or bandwidth?

What is the type of your offsite backup repo?
If you use a backup repo with xfs or reFS, Fastclone will be used. Bandwidth will not be a problem then.
Is it possible, that you use an CIFS Share as a offsite Backup repo and no veeam backup server is installed there? It would explain the „health check consume a lot of bandwidth“ situation.

11 hours looks to me, that FastClone is not used and no Veeam backup repo Server is installed in the offsite location. Then, when the synthetic full is running, every block from your 1.7TB data needs to be read from the storage and written back as a new full backup file. Over a slow WAN Link or VPN (approx. 5-8 mbit/s in your case?), this will be a pain for the network

[leockw]

Dear Mildur,

Yes, you hit my painpoint. We are using a Windows server as VBR server and primary repo. and NAS with SMB as offsite repo. without more VBR server.

Is that means if I have one more VBR server at offsite, the backup file health check and GFS synthetic full can do it at offsite directly?

Could you let me know if I build one more VBR server at offsite, how to config it and solve this bandwidth consuming problem?

Kind Regards,
Leo

[Mildur]

You need a simple Windows Server or Client at the offsite location. You don't have to install an entire VBR Server. Then you add this client/server as a managed server and configure it as a CIFS Gateway. This way, the Synthetic Full (and health Check too) will happen at the offsite location. Without it, your vbr server needs to read the data over the WAN Link

https://helpcenter.veeam.com/docs/backu ... ml?ver=110

A machine that performs the role of a gateway server must meet the following requirements:
- The machine must meet the system requirements. For more information, see System Requirements.
- The gateway server can run on a physical or virtual machine.
- The gateway server can run on a Microsoft Windows machine.
- You must add the machine to the Veeam Backup & Replication console as a managed server.
- The machine must have access to the backup repository — shared folder, Dell EMC Data Domain or HPE StoreOnce.

_______________________________

https://helpcenter.veeam.com/docs/backu ... ml?ver=110

If you plan to move VM data to an offsite SMB repository over a WAN link, it is recommended that you deploy an additional gateway server in the remote site, closer to the SMB repository.

_______________________________

Adding a managed client/server to the veeam installation:
https://helpcenter.veeam.com/docs/backu ... ml?ver=110

_______________________________

The gateway Server is then configured in the Backup Repo Properties:
https://helpcenter.veeam.com/docs/backu ... er=110#smb

[leockw]

Dear Mildur,

Many thanks for your infomation sharing.

When I add the gateway server, is it mandatory to add the shared folder? Do we need to amend the backup copy job to use the gateway server?

Kind Regards,
Leo

[Mildur]

You don't have to change anything in the backup copy job or add the shared folder to the managed server.
The managed server uses the configured authentification settings and Share UNC Path which is already configured in the Backup Properties.

1) Add managed server
2) Change the existing backup repo properties from your CIFS Repo to use the newly added managed server as a gateway.


Then, Backup Copy Job will run like before. Only the traffic will go once over your WAN Link. Incremental data needs to be send to the gateway server, the gateway server will then write the data to the CIFS Share. On the day, the synthetic full happens, veeam will send the incremental date through the gateway to the cifs share. After that, the old backup blocks will be read from the cifs share on the gateway server and written back to the cifs share as a full vbk file.

[leockw]

Dear Mildur,

Thanks a lot for your help and solve my problem.

However, do you have any idea on the process between the VBR server, gateway server and offsite repo when there is no any job is running? Since, I found the gateway server has over 100~200Mb sent out traffic when there is no any job running.

Kind Regards,
Leo

[foggy]

Hi Leo, are you sure it is a Veeam B&R-related traffic?

[leockw]

Dear foggy,

You are rights, I'm not sure what is the traffic but it just act as another VBR server and the gateway server. At the said moment, it has no other job is running from our primary VBR server and itself.

On the other hand, it just has send out traffic but a little receive traffic on the NIC.

Kind Regards,
Leo

[PetrM]

Hi Leo,

Perhaps, you could monitor the traffic on gateway server with Wireshark, at least you'll be able to determine network nodes that are involved in the process, it might give a better idea of traffic source.

Thanks!

[leockw]

Dear PetrM,

Thanks a lot, I will check it later.

Kind Regards,
Leo