I see a lot of questions asking “Why should we use VMware VSAN?” First, let me give an overly simplistic explanation of what a Virtual SAN does. Virtual SAN uses hard drives or SSDs in group of servers to create a group of storage volumes. Imagine a RAID array across different servers. Redundancy and distribution of resources can yield up to 99.999% uptime, and there are tunable failure tolerance settings. Add in snapshots, compression, deduplication, inter-site replication, and SSD caching for I/O performance, and VMware Virtual SAN is a real and valid SAN solution.
The first reason to choose Virtual SAN is that it costs approximately 50% the cost of traditional SAN implementation. Fibre channel switches and HBAs and switches are no longer required, although ethernet to the servers of at least 10Gbps is recommended. Proprietary storage arrays are no longer needed, so hardware costs are much lower than typical SAN implementations. Licensing with many SAN solution software solutions is also based on features or the amount of storage needed. VSAN is licensed per CPU socket.
Use your storage
By allowing the Virtual SAN to use the direct attached storage in each server, local server storage is not wasted. Without VSAN, when shared storage is needed for high availability, local storage on the servers is often wasted. Then, without VSAN, there are times when one host has most of its storage free, while another host is 95% full. With Virtual SAN, this becomes a volume of shared storage.
Flexibility and scaling
A big reason to choose Virtual SAN is that it scales well horizontally out to as many as 64 hosts per cluster and scales up above 8PB. IOPS can exceed 100,000 per host with flash storage. As a part of the software-defined datacenter (SDDC), the hosts are used for hosting VMs as well as the VSAN. This cuts down dramatically on hardware expenditures for a dedicated SAN.