VEB and remote sites

[Pat490]

I have one remote site with slow connection. Now I want to backup physical server from remote site with VEB to my main B&R repository. Can I do my first initial VEB backup e.g. to a local USB drive, than copy USB drive to my repository and after continue to incremental backup with VEB over the slow WAN connection?

[Dima P.]

Hi Patrik,

Backup file mapping is not supported for backup files created in locations other than Veeam backup repository.

Unfortunately, that’s the known limitation of the existing release. I hope we will be able to fix that in the next versions.

As a workaround, you may create a local repository at remote site, create a full backup there and then copy such backup to the main site since backup stored in the repository can be mapped to another repository

[Pat490]

ok. thanks for the answer. Is VEB always forever incremental backup? only 1 time full?

[Dima P.]

Yup, VEB uses forever forward incremental approach.

[Pat490]

ok over the weekend I did backup the remote server. It took 30hours but it worked.
strange thing is that it took 12 retries before the backup finally started!

the email notification shows always "The specified host is unknown. Job has been stopped..." (my own translation). which host does it mean? VBR server or target repository? which logs can I check for further investigation?

my other 2 remote servers did backup without any error to the same repository before.

[Dima P.]

strange thing is that it took 12 retries before the backup finally started!

This might happen due connection drops

which host does it mean? VBR server or target repository? which logs can I check for further investigation?

Open a support case from the affected VEB machine (VEB Control Panel > Support > Technical Support) and post the case Id to this thread please. I'll ask QA team to check VEB logs to identify the root cause.

[Pat490]

case opened with #01676578

[Pat490]

I had just been informed that our service provider had a power outage last weekend. that explains our problems at that time!

[Dima P.]

Patrick,

Thanks for the heads up. Anyway, it seems that we need to enhance the connectivity behavior between VEB and repository (make it more like the download manager to avoid long connection drops) – so thanks for submitting the logs.

[jhh]

Have now been testing VEB on a couple of remote sites backing up over WAN links (20 Mbit/s) to our existing Veeam Backup and Replication repository in our datacenter. File level backup mode is used. It works but takes forever even if there are only small or none incremental changes of backup data.

I use network throttling and limit it to 18 Mbit/s during night time and 5 Mbit/s during day/workhours and it works ok when transferring backup data. Then VEB starts with some kind of file scanning/inventory of backup files on the repository which saturate the links completely and takes forever since it seems to scan through all backup data files.

Example of backup with no new or changed data :


Found this thread http://forums.veeam.com/post152272.html ... st#p152272 which describes the same issue as Problem 1 and the solution seems to be to create a local Repository which I think is a bad workaround. If we were to use a local repository we would first need to invest in new backup storage on every site and then create backup copy jobs which has the disadvantage of losing all security groups on backup files if used with VEB backups if I understood it correctly.
Would really like to use only one backup solution for both virtual and physical but in this type of scenario VEB does not meet my expectations or is there a solution to this except from creating local repositories?

[Dima P.]

Veeam Endpoint Backup Free is not designed for a backup over the slow WAN links – that is true. Keeping this mind, I do see that described behavior is far from expected, so please open a case thru Control Panel > Technical Support and post case ID to this thread. Thanks.

[Pat490]

how about network traffic anyway? when I check traffic stats for the wan connection there is still a lot of traffic on 2nd incremental backup. But more incoming than outgoing to B&R Repository? Does it have to transfer first the full backup (.vbk) back to remote server to check what is changed and than only transfer new incremental .vib?



I do not understand the long time incoming traffic to remote site during last incremental backup (green)

[Dima P.]

Hello Patrick,

To clarify, is that a resource monitor screenshot or some router/firewall statistics?

[Pat490]

router statistics

[Dima P.]

Thanks. VEB job should receive the info about the exising blocks at the repository that is why you see the incoming traffic at the router monitor. The size of these digests is not comparable with the full backup size, so we are talking about several megabytes here.

[rodo]

Would really like to use only one backup solution for both virtual and physical but in this type of scenario VEB does not meet my expectations or is there a solution to this except from creating local repositories?

Create a local repository for Backup job and then create Backup copy job in VBR console to copy backup from local repository to main repository in HQ. We use this solution for 30 branch office servers.

[Pat490]

Thanks. VEB job should receive the info about the exising blocks at the repository that is why you see the incoming traffic at the router monitor. The size of these digests is not comparable with the full backup size, so we are talking about several megabytes here.

hmm I thought so but the incoming traffic looks more like several gigabytes. I will check again after the next weekly backup...

[Dima P.]

Thanks Patrick – I’ll be glad to hear the results.

[Pat490]

same result with backup on last weekend much more incoming than outgoing traffic!

[Pat490]

last night I saw the exact same behavior on another remote VEB endpoint. It looks like it first has to read/download the full backup from B&R repository and than uploads the new incremental!

[Dima P.]

Hi Patrick,

Your highest incoming traffic per second was 2040 x 0,000125 = 0.255 Megabyte. The peak was between 21 – 18 seconds, so it gives us somewhat 0.765 Megabytes of incoming traffic overall. Of course, if I am reading your network stats correctly.

[Pat490]

the bottom scale is time. so the incoming traffic was 6pm to 9pm! 3 hours maxing the line.

[Dima P.]

Patrick,

Thanks - now I get it. Please open a support case and let the team check your logs.

[Pat490]

I opened a support case with # 01700169

[jhh]

Have been testing some more now and switching to "Entire Computer" from "File level backup" mode solved my issue with long backup time over my 20 Mbit/s WAN links. Time needed for backup decreased from on average 8 hours to a very nice 45 minute backup for my server with total data of 400 GB (very low data change rate). When i did "File Level" mode total data selected was 290 GB, VBK file has increased from 200 GB to 290 GB with Entire computer mode.

On the downside is of course more space is needed for backup files since we now backup irrelevant data but disk is cheap compared to bying more bandwidth.

So my conclusion is select Entire Computer mode (have not tested Volume level yet) on remote sites if you plan to backup over slow WAN links.

[Pat490]

I opened a support case with # 01700169

my (short) reply from support is:
CBT and checksum of transmitted blocks make a network traffic.in your infrastructure, this is behavior by design

[Dima P.]

Patrick,

Thinking of Johan post above - are using the volume level backup or file level backup (sorry if it was previously asked, cant find any info for some reason) ?

[Dima P.]

UPDATE: We checked your logs from support case - the upcoming VEB update will have the performance engine improvements and that should correct the existing behavior. Stay tuned for update.

[Pat490]

we are using file level backup.

and thanks for the update. looking forward to the update...

[Pat490]

I still see the same behavior with VEB 1.5

Job duration: 2 hours
Size: 5 GB
Read: 45 MB
Transfer: 45 MB

but outgoing traffic on the connection line from repository to endpoint is over the whole 2 hours of the job!
incoming traffic for write is only the last couple of minutes.

So I still think VEB first have to read (download) the older backups from repo on local machine and than compares and writes incremental back!?