Backup / Replication Desgin

[BHB3805]

I currently am using VBR 6.5 to backup my VMware 4.1 U3 environment.
We have an intiative to begin virtual machine replication to our DR site.
The DR Site is connected via a dedicated current using a new VLAN.
I want to be able to perform replication at 2 - 4 hour intervals and maintain my backups (once daily, synthetic full on Fridays, 14 restore points).
What is the best method to accomplish this?
Multihomed proxy servers having interfaces for the ESXi management VLAN and the replication VLAN?
How do keep backup or replication jobs from locking files and failing the job.

[dellock6]

On 6.5 there is no problem for concurrency, if a VM is already locked in a job, the other job is queued regardless beeing another backup or a replication one, it will simply wait for the previous job to complete.
About network, I see no problem in having a multi-homed proxy server, only be careful that if the Veeam Central server is on another server than the local proxy, it would need to be able to connect to both local and remote proxy, so you would also need some routing between vlans...

Luca.

[BHB3805]

The replication link is designed to be a non-routeable circuit between the production and DR site.
My thought was to have a proxy (VM) in the Production site with two vnics - one to the production network and one to the replication network.
I would do the same on the DR site.
I can perform a Windows File copy between the VMs, but I'm not sure how to force Veeam to use the replication link for replication.
Any suggestions?

[dellock6]

Uhm, how can you register the remote proxy into Veeam console if that IP is not routed? I suspect by using the production IP, but my suspect is that in this way proxies will see each other via those addresses. I had a case where customer wanted a completely isolated backup network, this is not simply possible since proxies need to connect to ESXi servers via some tcp ports, as you can see in this kb article:

http://www.veeam.com/kb1518

Luca.

[BHB3805]

I registered each proxy using it's Production Network
Here's what I have:
PRODUCTION
ESXi 4.1 U3 - 10.4.68.10/24 GW 10.4.68.1
vCenter - 10.1.48.10/24 GW 10.1.48.1
VBR 6.5 - 10.4.68.20/24 GW 10.4.68.1
VBR 6.5 Proxy - 10.4.68.30 GW 10.4.68.1 and 10.1.76.30 No GW
DR
ESXi 4.1 U3 - 10.1.32.90/24 GW 10.1.32.1
vCenter - 10.1.32.10/24 GW 10.1.32.1
VBR 6.5 Proxy - 10.1.32.30 GW 10.1.32.1 and 10.1.76.30 No GW

[foggy]

I can perform a Windows File copy between the VMs, but I'm not sure how to force Veeam to use the replication link for replication.
Any suggestions?

Wouldn't the Tom Sightler's suggestion to put static routes on the proxy servers to route to each others production interface via the replication interface be helpful here? I take the liberty to quote Tom's post:

OK, so here it goes:

Network-10g.png (55.65 KiB) Viewed 11229 times

So in this example we have two networks as follows:

Management (1Gb): 10.10.1.0/24
Data(10Gb): 10.10.2.0/24

The problem is, the Veeam server will know the two servers by their management IP address/DNS name and likely be completely unaware of the 10Gb network at all, so, when it starts a backup, it will instruct the repository at 10.10.1.20, to connect to the proxy at 10.10.1.10 which means the data will flow over the 1Gb network by default. However, we can change this behavior as follows:

On the Wintel proxy add a static route with the following command (the -p makes the route persistent across reboots):

Code: Select all

route -p add 10.10.1.20 mask 255.255.255.255 10.10.2.20

On the Linux box add a static route as follows (method to make this persistent across reboots varies based on distro, although I guess you could just add it to rc.local):

Code: Select all

route add -host 10.10.1.10 gw 10.10.2.10

So Veeam will still tell the repository at 10.10.1.20 to connect to the proxy at 10.10.1.10, which is the 1Gb interface, however, according to the routing table, the way for the repository to reach the proxy at 10.10.1.10 is via the "gateway" at 10.10.2.10 which is the 10Gb interface of the proxy itself. The reverses is true for proxy to repository communications. With these two route entries, even though the proxy and repo will still be using their "management" IP addresses to establish communications with each other, they will route all traffic to each other (because the routes are "host" based, not network) via the 10Gb interface.

The most elegant way to implement this (as far as I understand, this is similar to your case).

The other way is to register both proxies in Veeam using their DNS names (which should resolve to their production IP's) and use hosts file on each of them to point the counterpart's DNS name to its IP address on replication network.

[Cragdoo]

Does anyone have any experience of using multiple IPs/vlans on backup/replication proxies,and creating dedicated replication links ?

Here's the scenario :- (IPs are for simplicity sake)

Veeam backup server on IP 192.168.1.10 (vlan 192)
Backup proxy on IP 192.168.1.20 9 (vlan 192)
DR proxy on IP 192.168.1.30 (vlan192)

Both the Backup Proxy and DR proxies have additional interfaces (10.0.0.20 and 10.0.0.30, both vlan 10) respectively.

Is it possible to configure a replication job that will use VLAN 10 for the replication data transfer ?

We're trying to setup a remote replication and create a separate replication vlan (vlan 10) and management vlan (vlan 192) over the same link to the remote site.

Apologies if this has been discussed elsewhere , please feel free to direct me to the appropriate posts.

[PedroVDP]

I have two servers, which both were upgraded to v7 which serve as 2 independent silo's in two separate datacentres.

These servers are connected on both an administration VLAN and a storage VLAN. Both these VLANs are L2 extended across the datacenters but using separate network lines.

How could I proceed best to force the connection between both proxies on those servers to use a specific network interface? I'd prefer to run Backup Copy traffic over the storage VLAN.

[foggy]

Pedro, please review this thread, should give you some hints.

[Cragdoo]

Sorry for resurrecting an old thread, but I have tried both methods on a V7 replication setup , all routing and DNS points to the 'replication' interface/link but the replication data is still moved over the 'management' interface/link. Has anyone ran into similar issue with V7.

If I carry out a network file transfer between the 2 servers, the 'replication' link is used, but the replication job still uses the 'management' link.

We have/had a similar setup on a V6.5 infrastructure and it worked as expected, I'll need to check this is still the case after it was upgraded to v7. Sound like the DNS lookups are performed elsewhere in v7 ...

[fast_cat]

Due to some network bandwidth limitations at a particular site I am looking to send ALL backup data over a dedicated backup network to keep the traffic clear of the production network.
At present I am testing with a veeam proxy which is multihomed and a repository server which is also multihomed. The Veeam Backup server is not multihomed.

I need the proxy server to send its backup data (and restore) over ht ededicated backup network and not the primary. However the primary interface is what is configured on the backup server since it has no routew to the backup network.

I have tried with a local dns alias on the proxy but backup traffic is still being sent over its primary interface.

Does this make sense ? If so how is it possible to direct the backup traffic ?

[danielchung]

I'm making the similar setup in my test lab and found this post.

Here is my setup:
- Veeam B&R server v8 (vlan 10)
- vCenter (vlan 10)
- ESXi host (vlan 10)
- multihomed backup proxy (vlan 10 and 192)
- multihomed repository (vlan 10 and 192)

The vlan 192 is dedicated backup network, only the backup proxy and repository servers can reach vlan 192. I added the proxy and repository servers into Veeam B&R server using their DNS name, and also edited host file and sure that IP addresses are resolved correctly (vlan 192) via Ping. However, the backup data still being transmitted over the vlan 10. It seems like the Veeam server resolved the IP and told the proxy and repository servers, instead of resolving by themselves...

Anyone got it working?

[alanbolte]

For proxy-to-repository traffic, you can specify preferred networks in v8:
http://helpcenter.veeam.com/backup/80/v ... twork.html

[Liro10]

At my production env I have 2 network cards, one is for production traffic and one for backup only,
All my env based on vmware and windows server 2008r2
On the VMware level, all backup and production traffic are physically separate NICs.
Once I install the Veaam to test the program I didn’t find any way to force the backup to run under the Backup card only after few resarch i found way to do that (Networks => Preferred network) => but still dosen't help...
i am still see that the backup is running only on our production card and not on the backup card Originally I've discovered the VMs via 2 * VCenters, but the discovery was done using the production network as the Vcenter VM management network is *not* the backup network. Is it possible that due to the discovery of the VMS was done over the production and not the backup network that I am seeing this phenomenon?
Is there possibly something I can change on the Veeam client installed on each VM I backup regarding the backup IP address?
Is there is any way to configure or change the ip's for the backup process?

Any suggestion?

[foggy]

Are backup proxies added to Veeam B&R console using addresses on the backup network?

[Liro10]

Yes, i also tried that.... my Main network is 10.195.20.x
my backup network is 10.195.21.x
i created the veeam to use proxy (10.195.21.29) and still i can see that both network cards are working, and the main network work much more harder than the backup network..btw- i don't have dns for the backup network ip's..

[foggy]

How about your repository? How it is added to Veeam B&R console and what address does it have?

[skrause]

If I understand what you are trying to do, it sounds like you will need to set static routes for the second network in Windows. I had to work with something similar for a plant monitoring system in a previous job where they were on two separate networks that were not routable.

I don't remember the specifics, but I do know that I relied on this article https://technet.microsoft.com/en-us/lib ... 10%29.aspx .

[Liro10]

the repository using the backup network \\10.195.21.x => i already think about that..

Skrause- you think this is the only way?