Copy to Cloud: SOBR with Capacity or Cloud Connect for the Enterprise ?

[TinchoB]

Hello,
as our backup appliance we have a B&R v10 deployed on premise in a physical server attached to a SAN Storage (standard repository). Our repo has 106TB in sieze and is using NTFS. We are doing an active Full backup per week (Saturday) and then one incremental per day. We use per VM processing. No GFS jobs set. Our retention is 30 restore points (one per day). The size of a weekly full backup is about 10TB (after VEEAM deduplication/compression). We use VUL Essential Subscription (so using Enterprise+).

Now, we have to expand our solution to comply with 3-2-1 rule (backup copy externalization) and also enable some kind of history (with monthly/yearly restore points). Our objective is to have at least 8 weekly, 13 monthly and 3 yearly.

Our management ask us to make an exercise to determine the best option from the cost/performance/operational point of view between:
Option 1 (add a new Tape Library with X amount of tapes and extract and rotate one media set per week)

or

Option 2 (copy to Azure) (we currently have a 150Mbs all purpose internet up-link in our site, but no Azure account contracted).
Regarding option 2 (copy to cloud), according what I could find on internet there are two approaches:

A- Contract an Azure Blob Storage (cold) and use it as a Capacity Tier in a SOBR (using the present repo as the Performance Tier) and play with copy/move policy. Most of the references I could find talk about this way.

B- But I also read about other way based on Cloud Connect for the Enterprise, were we should deploy a specialized VM from the Azure Marketplace and then attach some kind of Azure Storage to it. And then establish a connection with our site and define it as a Cloud Repository in our on premise VEEAM Server, and play with backup copy jobs.


Could someone please help me to determine if its correct and what is the use cases for every approach (1, 2A, 2B) or which one do you think could be the best for us ?
Is Cloud Connect still valid for our approach or is it better to go with SOBR ?

Also, what is your experience with costs and cloud storage ? What should be the less cost one ?

And what for recovery times form Cloud ?

Thanks you in Advance !!

Tincho

[Gostev]

Hello!

You should go with either 1A or 2. Tape storage will require CAPEX, but will always be much cheaper in the longer run no doubt. Although keep in mind you'd need to deal with it physically - so you should consider personnel costs too.

Capacity Tier "just works" requiring no on-going management, but the storage is more costly, especially with Azure (which has its object storage priced at a premium). The other concern here is your bandwidth (especially in case you need to restore a lot of data from the cloud).

Cloud Connect for the Enterprise has a completely different use case and it is not really applicable much to your scenario. Also, it is going to be the most expensive option if you host it in Azure, as opposed to in your own data center.

[soncscy]

Heya Tincho,

>Option 2 (copy to Azure) (we currently have a 150Mbs all purpose internet up-link in our site, but no Azure account contracted).

This seems challenging with your data amount, unless you seed it with Azure Databox: https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Without this, even with the link fully engaged, you'd need 6.4 uninterrupted days, which sounds tricky for me. After that, you just need to size out the time to transfer the incremental data size, as the way Capacity Tier works should keep the offload "petite". I always just use this simple calculator for such things: http://calctool.org/CALC/prof/computing/transfer_time

So, just be aware that special accommodations will need to be made if you end up starting new backup chains for some reason as you'll need to "re-seed", but it's an option at least for your backups.

I kind of like tape in this situation personally as I'm guessing your 150 link is shared for other uses also, and I just don't think it's really good for the amount of data you'll be moving weekly, even with the WAN savings that Capacity Tier gives, but that's just my mind here.

[TinchoB]

Hi Gostev and Soncsy,
thanks four your answer, I will take them into account (Tape: physical write protection, faster in the event of a disaster, less cost but manual operations. SOBR with Azure: no manual handling, but slower due to internet bandwidth limitations and costly).

In order to compare Tape Library Option (with backup copy job to tape) to SOBR with Capacity on Azure (with original backup JOB with GFS enabled), I should try to estimate costs values.
According what I could read in the forum, capacity tier offloading is "like" a forever incremental, so I could use Restore Point Simulator selecting BackupCopyJob GFS using ReFS, right ? This will give me an amount of required space. Then, as you pointed out, there is the issue of the internet bandwidth, which in case we go by this way should we have to increase.

Then I have another question.
Suppose we go with SOBR Capacity tier in Azure Blob, and in an emergency (or for some other reason) we have to restore a VM but direct in Azure (to avoid our internet link limitation... or because the main site is down), should we also deploy a VEEAM BACKUP for AZURE server in the cloud (just in case), connected by some way to the backups in the Blob storage ? Could we have the VEEAM on-prem Server and the Azure one accessing the same blob ?

I´ve to give it a think, but perhaps we can have some kind of combination between both solutions: most in tape, and some specific things in the cloud.

@Gostev, which one should be the case use for Cloud Connect for the Enterprise ?

Thanks !!

[Gostev]

In a word, multi-tenancy. The main use case for Veeam Cloud Connect for the Enterprise are ROBO environments that have many remote offices and remote workers who need to backup securely back to the main data center. Unless you have at least a few dozens of such separate "tenants", the management overhead of maintaining a Cloud Connect infrastructure just does not make sense from the TCO perspective. You just put a simple VPN in place instead and be done with it.

[veremin]

Suppose we go with SOBR Capacity tier in Azure Blob, and in an emergency (or for some other reason) we have to restore a VM but direct in Azure (to avoid our internet link limitation... or because the main site is down), should we also deploy a VEEAM BACKUP for AZURE server in the cloud (just in case), connected by some way to the backups in the Blob storage ?

If local backup server is still running, you can use it to restore backups from object storage directly to public cloud - no issues here.

Otherwise you will need to deploy new backup server (can be local or cloud one - does not matter), add previoгsly used object storage repository, import backups from it and start restoring VMs.

[TinchoB]

veremin,
Does "direct" restore from object storage to Azure means that no data has to be download on-prem to be then uploaded to Azure, not ?
Thanks !

[TinchoB]

veremin, do we also need to have an Azure Proxy deployed ?

[veremin]

Does "direct" restore from object storage to Azure means that no data has to be download on-prem to be then uploaded to Azure, not?

Correct.

veremin, do we also need to have an Azure Proxy deployed ?

Correct.