Plannification between job...

[Stabz]

Hi guys,

I have two production sites (SiteA and SiteB), SiteA is the DR site for SiteB and vice versa.
On each site I have this configuration:
-Production Esxi cluster (v5.5) connected to a VNX5300 in FC.
-A standalone Esxi(v5.5) which is dedicated for the backups and restaurations. There is local storage in it and we have direct attached a MD1400.
The MD1400 is used to store the backup and the local for the replication.
This Esxi have also a link to the VN5300 in fiber channel

Currently we have backup jobs every night and we export them in the other site with backup copy jobs.
Everything works fine, but we changed our RPO/RTO for some applications servers, so now I have to implement replication job.

We have a 100 MBits MPLS link between SiteA and SiteB.

I would like to setup a replication job between my production and backup in local, and have a second replication job to my DR site.What would be the most effective way to do this?
Of course I want to preserve my backup and backup copy jobs.
Another point, my VNX5300 don’t have the module for Storage Snapshot.
So after multiple headaches, about how to plan the different job and avoid overlapping between backup jobs, replication jobs, backup copy jobs…., and reduce the number of snapshot on my production, here is a first approach:
- Create a new backup job with the good RPO
- Create a replication job from backup files
- Create a backup copy to my DR site
By doing this I have only one snapshot on my production, but if my backup is corrupt my replication will be too…
So do you have good feedback to do replication from backup?
And I still have to replicate this Vm on my DR site….

I’m a bit confused about how to apprehend this project.
All help is welcome!
Philippe

[PTide]

Hi,

Did I get it right:

- all backup infrastructure components (proxies, repositories, VBR) reside on VMs on the standalone host
- backups are saved to MD1400 LUN. Is it NTFS volume over VMFS, or just NTFS provided to a virtual proxy via RDM, or is it smth else?
- you use hot-add transport mode

Is everything correct?

[Stabz]

Hi PTide,

1) Yes, all backup infrastructure components reside on Vms on the standalone host.
2) It s NTFS volume over VMFS, I never try to work with RDM,perhaps wrongly.
3) My VBR has in this configuration, two PCI Device which correspond to my FC card, so I can use Direct San or Hot Add

[Deon]

but if my backup is corrupt my replication will be too…

Hello Philippe, could we focus a bit on this worry of yours?

If the backup file is physically corrupt, it will trigger an error during the read operation by replication from the backup, so you should receive an email notification at once and be able to deal with that.

Is there some other "unnoticed" corruption you are worried about and what did you consider? The only issue I am thinking about is an actual data corruption on volume level (in guest OS) because of CBT issues, but that's very rare and I think most of such issues were patched on latest releases of ESXi 5.5. And quite often it's possible to fix this issue by disk repair utilities.

If you share what you think about it, we may find a better idea.

[Stabz]

Hi Deon,

Thanks for your answer.

I don't have a precise idea, like I said , I never try this feature so, I prefer have feedback before implementing it. But according your statement is a viable solution.
I know that I don't have a lot of possibilities with my configuration, execpt doing many snapshot but I'm not a big fan of this idea..

In my offsite, could I create a replication job from backup copy job or not?

[PTide]

It is possible, however the implementation will depend on the number of VBR servers. Do you have one or two servers?

Thanks.

[Stabz]

Ok!

My plannification will be :
-Backup my critical vms every: 1h
-Replication from backup when backup job is finnished
-Backup copy job to offsite
-Replication offsite from the backup

Seems it logical ? I'm a bit affraid to do a little to much

[veremin]

How many backup servers you have? One, right? In this case there should not be any issues.

By the way, I'm getting it right that you will be using two different sites as replication targets: onsite (replication from backup) and offsite ones (replication from backup copy)?

Thanks.

[Stabz]

Hi Eremin,

No I have 2 backup servers, one in each site.

But yes I need to replicate onsite and offsite. I agree, it's a unwieldy process.

Replication onsite from backup seems to work properly, but now how to implement replication from backup copy.
On my DR site:
- I added the backup repository of my backup copy
- I created my replication job, in source I checkec the repository
- I configured the rest of the job

But when I start it, I obtain the message "05/10/2017 15:30:14 :: No restore points found"

[veremin]

You will also need to create a dummy backup job, map it to files created by backup copy job and also schedule a periodic repository rescan that should happen prior to replication job (search through these forums, there must be several examples of people doing exactly the same). Thanks.

[Stabz]

Hello Eremin,

Ok, I will take a look at this solution, but in my memory this solution requires a lot of manual work and it's just a workaround and not a real solution right?

[veremin]

Correct, but given the complexity of your scenario, it will be always a workaround, not a solution. It stands to reason to consider having only one backup server and orchestrating everything from it - the complexity will decrease, and no manual efforts will be required. Thanks.

[Stabz]

Yes, I understand...
To follow your idea, in my case, each site will manage all the jobs (backup,backup copy, replication) for the distant site.

I have always been cautious with this type of configuration in the past, because when I tried it to do a restore backupfiles must transit twice (in my memorie) on the network...

So my solutions are:
- Doing a tricky configuration with some workarounds....
- Implement onsite replication from backup and offsite replication from infrastructure, but there will be two snapshots,
- Ask for an investisment in VNX5300 snapshot licence
- Or try to manage everything from a VBR in the DR site

[veremin]

- Or try to manage everything from a VBR in the DR site

If I were you, I would probably stick to this option. DR backup server and components locally (proxy, repository) deployed locally should be more than enough to address every need that you have. Thanks.

[Stabz]

Ok, I ll try it next week in my lab and check if this configuration meets my needs.

Thanks

[skrause]

Yes, I understand...
To follow your idea, in my case, each site will manage all the jobs (backup,backup copy, replication) for the distant site.

I have always been cautious with this type of configuration in the past, because when I tried it to do a restore backupfiles must transit twice (in my memorie) on the network...

You can now set a mount server in B&R for restores that removes the cross-site transit. The only traffic that needs to go between sites is control traffic, just like in a backup job.

I orchestrate all of my backup jobs for multiple sites from one remote B&R server and it works well. (I just set the mount server for each Veeam repository to the server that hosts said repo)

[Stabz]

Hi Skrause, thanks for your feedback.

Im testing this architecture in my lab.

Backup and Backup copy seems to work properly from my DR site.

But I 'm facing a "problem" in my replication job.
The calculating digest disk is very slow...

This is a replication onsite. The configuration is :
- Production Vm source my backup
- Repository for replica metadata is the same as my backup repository
- The target proxy is my backup repository too

After 20 minutes for a disk of 80Gb , I have only 4% completed....

Any idea?

[PTide]

Any idea?

So, now you have the job that replicates from the backups on the PROD site, while the VBR is on DR, both target and source proxies are on the repository VM where the source backups reside, and the target replica ESXi host is on the very same PROD site. What are the bottleneck stats?

Thanks

[Stabz]

Yes tha's it!
I tried to change the proxy, but the problem remains the same..
The bottleneck stats don't give any informations the status is "Detecting..."

I tried to do a replication from the infrastructure and it works well.

[PTide]

Please double-check that the replication and backup jobs do not overlap. Also I'd check the performance of both proxies in order to find out if it needs more RAM or CPU. If nothing helps then please contact our support team directly so they can take a closer look.

Thanks

[Stabz]

Indeed there was overlapping between the jobs.
So I disabled the backup job for the moment, but the calculating task still slow... I'll wait until the end

I already checked the performance of VBR, Proxy and repository servers, but nothing is anormal...

I ll open a support case

[Stabz]

Hi guys,
A little update

I waited the end of the calculating digest, after that the job worked well, until yesterday ....
I don't know why, but the task calculating digest is appeared and was really slow... So I had to disable my backup job to avoid any overlapping job...

I opened a case 02342732, I ll do a feedback once I receive more information from the support.