Second opinion needed for backup solution

[gstarr]

Hi,

I really love veeam,use it to for backup and for replication to offsite location.
Now I have a big problem.
Our company has third location in another country.

There are 3 servers - oracle, application and DC.
All physical, that means that veeam is out of the question.
I have been searching for reliable physical backup with features on par with veeam... no luck.

Since this is a beginning of operations in that country, big IT investment for storage is out of the question until company grows and expands a bit.
In order to use veeam, I have an idea to install vmware vsphere essentials on each of this servers.
That means that each host, would be a host to only one virtual server.
In a way similar to the current state, only that I would have benefit from veeam, which means much faster
restore in case of server failure plus replication to our primary location.

Ideally use of storage or storage appliance would be beneficial but would increase the cost.
So currently this is a no go.
Did any one else had case similar to that one?

[steve21]

what you could use is Symantic's Back-up Exec. i have been using the 2010 version with the combination of veeam to back-up my servers for a while now and it has been working just fine for what i need it to do. i know with the 2012 version of back-up exec there have been many people that have had problems with it only because it is different that the 2010 version.

[gstarr]

I do use Be2010 now.
you dont have replication and you dont have easy restore.
with veeam there I could restore in a minute or in case disaster I
could run the servers from main location.

[dellock6]

Hi, following your opening post, maybe you can consolidate the 3 physical servers in less than 3 ESXi hosts? If for example you can run all the servers in a single ESXi host, the budget would be limited to an additional Veeam license and a VMware license to enable VADP libraries on it. One of the other servers could become a spare ESXi server to be used for local replica (via Veeam) so you can also have a local DR solution.

Luca.

[Yuki]

there is a number of things you could try, but all of them come down to one of the following:

1) having storage device at at least one location
2) physical servers having enough space for at least two servers


you could setup a replicatio triangle where every server hosts just one VM but then each Vm is replicated to one other server. That way you have one active VM and one replica on each ESXi host. You could also set replica retention to about 14-30 days and use replicas instead of a backup (if you don't have dedicated storage device). It's not the cleanest solution, but an option.

if you can get a storage such as inexpensive NAS or even convert one of the server into a file server (NFS or iSCSI) you could use it as a backup repository.

To be honest, if your budget so limited that you can't afford storage - Veeam is probably not the right solution.

[zoltank]

Hi, following your opening post, maybe you can consolidate the 3 physical servers in less than 3 ESXi hosts? If for example you can run all the servers in a single ESXi host, the budget would be limited to an additional Veeam license and a VMware license to enable VADP libraries on it. One of the other servers could become a spare ESXi server to be used for local replica (via Veeam) so you can also have a local DR solution.

Luca.

This is exactly what I would do, and I've already done something similar to this. We had several servers in our DMZ which we were backed up using BE. We ended up buying one server, installing VMWare on it, and then consolidating all the server onto the one host. At least this way we could use Veeam to back them up, reduce our hardware costs, use snapshots during upgrades, and allow for very rapid recovery in case something happens. It's well worth it.

[gstarr]

The only problem with would be latency I guess and in case of failure - all of the servers would fail.
Does any one uses replication to another host?
does it work all right?

[Vitaliy S.]

If you're referring to replication over the WAN link, then it works just fine, even with high latency. Please search these forums if you want to get more info on that. Thanks!

[gstarr]

I was thinking about replicating to another host.
If I move all of the 3 physical servers to one host, use another one as a replica...
If host that hosts all of 3 servers fail, I have bigger problem than if only one physical server
fails. And by having to rely on server instead of 3, probability that it fails is higher than If using all of the 3.
If I replicate - local replica - all of them to the second host I am able to recover fast, question is how
good replication works in the local network, in time, If I restore replica, how much data would be lost
if you understand what I mean. Is that something that ammounts to minutes or hours?

I do have some small NAS which could be used for nightly backup. Over the day I could use local replica.

Another option would be that I virtualize only 2 main servers, and use both as a target for local replica of another server, so in case one fails
the other works, but I am able to restore the other faster.

I would leave 3rd one physical as it's just a DC, perhaps a copy of it as a turned off VM (converted via VMware convertor)
and In case that one fails, I just turn it on.

That way, I would have same functionality as with physical servers, but much better DR plan.
If one server fails, the other could be turned on and loss of data would amount max to one hour.
the same goes for second server.
And if physical server fails, I could bring up turned off DC which would be a copy of physical server.

[Vitaliy S.]

And by having to rely on server instead of 3, probability that it fails is higher than If using all of the 3.

It is, but having a standby replica should solve this problem. The question is - does your single host have enough CPU/RAM to support your three applications? If it is, then I would put all VMs on a single host and run replication jobs targeted to the spare host.

question is how good replication works in the local network

Works flawlessly

in time, If I restore replica, how much data would be lost if you understand what I mean. Is that something that ammounts to minutes or hours?

It depends on your replication job schedule. If you're going to run your replication jobs once an hour, then you will be able to failover to the previous hour restore point.

That way, I would have same functionality as with physical servers, but much better DR plan.

I would suggest going 100% virtual, as using rebuilding a physical server might take some time. Besides, you would have to run VMware Converter on regular basis, cause if you do it only once, your DC will be outdated.