I have a client that has about 20TB of data the needs backing up on a daily basis. Their environment consists of 3 physical servers and 4 Hyper-V VMs. One of the physical servers is configured as a virtual host while the other is the replication server. The VMs are configured to replicate every 15min. Both physical servers have 10GB NICs that are Teamed for load balancing.
We’ve installed two Synology DS918+ Plus NAS devices each with 8 x 4TB drives. Our idea here was that data would back up to the first NAS then we’d configure the secondary NAS as replication of the first, this second NAS would then be located off-site.
We’ve tried running the solutions on the VM host and have installed the software on each of the VMs to backup individually, each solution takes anything from 45 - 670 hrs to complete a single backup which is not at all productive. Our DR acceptable time is maximum to 15hours.
You can always consider paid recovery concepts as explained below-
Data center replication is the main concept behind disaster recovery. There are three main types of data center replication.
Hot Site: A hot site is a mirror of the data center infrastructure. The hot site has servers and infrastructure, which run, concurrently with the data center, syncing with it to provide complete redundancy.
Cold Site: A cold site is just datacenter space with no servers or equipment set up. In order to be operational, extensive engineering and IT assistance is required to install servers and other equipment and migrated data and software.
Warm Site: A warm site is somewhere between the two. It will have pre-installed servers ready for installation of production environments, but that work must be done before the site is operational.
Metrics
Two important metrics determining what kind of data replication to choose are:
RTO – Recovery time objective. The target time between the moment of the disaster and the point at which users can work again.
RPO – Recovery point objective. This is the age of files that must be recovered from backup storage for normal operations and business continuity.
Pricing Information
There are a variety of pricing models in this domain, including charging by each GB of storage used, an average per user cost, or a per device model. Typical pricing ranges from $10 per instance, per month, up to several times that for larger-scale enterprise products.
