The difference between VMware's VSAN and a VSA

There has been a lot of stir in the storage arena since VMware unveiled its Virtual SAN at VMworld 2013. Many have said it is a great product and many others have said that VMware should stay out of the storage market.

With VSAN now generally available, there are many questions about how VMware's VSAN is different from a traditional software-based virtual storage appliance (VSA).

A traditional VSA is generally a VM that runs one or more ESXi hosts in your environment. The host uses an existing data store or direct-attached storage (DAS) to provide VMDKs to the VSA for carving up into usable storage. The option to pass through raw DAS devices to the VSA is also possible if hardware permits this. Now, the natural question is, "Why would you do this?" This approach provides the ability to use capabilities provided by the software vendor in higher-priced NAS/SAN products at a fraction of the cost. These capabilities could include snapshots, replication and other important features.

VMware's VSAN is completely different and generally used for different reasons. VSAN is a product available only in vSphere 5.5 and is embedded into the vSphere kernel. VSAN has different requirements and provides a SAN approach that is enterprise-ready.

One of the requirements of VSAN is that every ESXi host providing storage must include a solid-state drive (SSD), which provides caching capabilities to the VSAN cluster. Not all ESXi hosts within the cluster need to provide storage in order to consume the storage provided by the VSAN cluster, but VMware recommends at least three ESXi hosts within the VSAN cluster providing storage and SSDs. This provides resiliency in case of a host failure on one of the ESXi hosts that is providing storage.

VSAN allows businesses to provision complete compute stacks including storage in a scale-out fashion. This provides the ability to grow as requirements change and removes many of the shortfalls associated with SAN-based storage. This approach can also bring the data closer to each workload, based on VSAN storage rules and profiles. It also gives you the ability to get a more granular approach to how a workload's data is stored, protected and cached. Having this ability is crucial as your requirements and capacity needs change, instead of having to buy blocks of overpriced SAN nodes and cages.