An Overview of Virtualization Storage Models and Architecture
Virtualization started off as a way to get more out of modern hardware. It was initially used to split disparate workloads and to isolate different users with different needs on the same physical hardware. In the “old days”, we had to but a new server if the finance department needed a dedicated server to run the month end budget on. We had to do a business case, purchase order and wait for the suppliers. This often took a few weeks. Now we can fire up a virtual machine in a few minutes.
Other Benefits of virtualization
These virtual machines are not only quick and easy to create, they are reasonably simple to “snapshot: and create a backup of. This does mean that you need a large amount of space on your backup server to cope with all the “snapshots”. Other benefits are that virtual machines can be moved around to physical machines that have a lower load or when a physical server needs a hardware replacement. These movements can be painless or painful. It all depends on your storage scenario.
The Storage Scenarios
Direct Local Storage
In the beginning, local storage was used extensively. This was fast and easy to maintain. The setup was simple and also not expensive. The downside of this is the “all the eggs in one basket” scenario. This is because you now have multiple servers on one physical host. If something were to go wrong with the physical host, the impact to business is so much greater. Virtualization with local storage would increase the risk profile in IT significantly. Impact of downtime on Business Continuity is far more extreme.
Shared Storage (NAS/SAN)
Shared storage was a great architecture for reducing risk. It created a built-in hardware redundancy as VM’s could often be live-migrated in case of a hardware failure. Doing it properly was expensive as your SAN or NAS storage has to be duplicated for redundancy. Network connectivity between your physical servers (Hosts) and the Storage layer has to be dedicated and as fast as the current technology allows. Currently, this means at least a 10GB/s copper or fibre backend. This drives the cost of SAN / NAS storage up. The set-up of a mirrored SAN or NAS is also quite complicated. Enterprise-grade SAN or NAS hardware can also very easily run into millions of rand. Advantages include quicker backups, often built-in deduplication and real-time failover (HA) for your virtual machines.
This is not, as some may think, a partial use of both local and SAN storage. This is local storage with SAN tech applied. You need the high-speed back-end connectivity, but not the expensive storage devices. How does it work? Well, the boffins have come up with a way to run all you local storage/disks as members of a networked cluster. This cluster then ensures that the data has at least 2 and usually more copies across the cluster for redundancy and failover. The data also has a copy on the physical machine where the VM is running. This gives you the speed of local storage without the risk. The cost of the SAN / NAS infrastructure is also greatly reduced. Hybrid storage is usually run under a specific software as “Software Defined Storage” or SDS. This software is usually proprietary and supported by the software vendor.
The Virtualization landscape is constantly evolving and at HOSTAFRICA, we are in permanent study mode to keep up with all the exciting developments. At the same time, we do our research and look to our sister company, Dogado, in Germany to see which solution works the best and give our customers the best experience and lowest risk.