In January, 2014, Wikibon, as part of its ongoing coverage of Software-led Infrastructure, provided a definition for a new architecture, they called Server SAN. Many look at Server SANs as the long-term disruptor to traditional datacenter storage arrays, but, just as importantly, organizations are beginning to use Server SAN solutions in remote offices and distributed enterprises.
For many organizations, software-only solutions such as StoreVirtual VSA from HP, SvSAN from StorMagic®, and VSAN from VMware, are the logical replacement options for RAID systems, SANs and NAS, in remote offices and distributed enterprises. In these environments, the storage capacity and performance requirements can typically be met with the internal storage capacity of branch-office servers. Just as importantly, Server SANs eliminate the single-point-of-failure concerns of a single shared RAID system. This is particularly important in a virtual-server implementation, where downtime in a RAID system will bring down with it all applications running on connected virtualized servers.
If the applications running in remote offices and distributed locations require high availability and the organization uses traditional RAID systems, then two arrays will be required. If they use a SAN, then they will require redundant switches and redundant arrays to eliminate single points of failure. Few organizations with large numbers of remote locations can afford this level of redundancy, nor is it advisable, unless the storage requirements in each location are large.
Action Item: With the advent of Server SANs and the increased use of virtual servers in branch offices and distributed enterprises, organizations should consider replacing traditional RAID systems with software-only Server SANs to eliminate single points of failure, gain the high-availability benefits of Server SANs, and avoid the cost and complexity of RAID systems and SANs.