Typical best practices for storage area network design involve each storage device connecting to a switch that then connects to the servers that need to access the data.  To ensure high availability of the data, we utilize a design that makes sure the path to and from the SAN isn't a single point of failure.  Each SAN controller or head unit should connect to both switches, as should each server.  If either path fails, multi-path software (MPIO) can fail over to the connection that is still active.  Depending on the desired results, we recommend configuring the host machines to use both paths if they're available, for load balancing.

A good storage area network design should not only accommodate your current storage needs, but it should also be scalable so that you can upgrade the SAN as needed throughout the expected life expectancy of the system.  You should consider how scalable the SAN is in terms of storage capacity, number of devices it supports, and  storage interface connectivity (FC, iSCSI, FCOE, Infiniband etc.).

Both of the SAN manufacturers that Enterprise Networks Solutions represents (Nimble Storage and Dell Compellent) offer a feature called thin provisioning.  Thin provisioning is way to give servers more pace than actually exists.  For example, an operating system (OS) that connects to a given LUN/Volume may think the LUN/Volume is 1 TB, even though you have only allocated 500 GB of physical storage for it.  This is very beneficial when building servers so that you don't have to repartition the drives when you are running out of space.  The only gotcha is, that you better keep an eye on the physical storage available on your SAN.  If you run out of storage space on the SAN, you can be in a world of pain with your servers.

For Disaster recovery there are two options:  synchronous or asynchronous replication.  Synchronous replication means that as data is written to the primary SAN, each change is sent to the secondary and must be acknowledged before the next write can happen.  Asynchronous replication does require the DR SAN to acknowledge it received the data before the next write can happen.  Each of these methods has pros and cons, but typically we use asynchronous replication.

These are a few of the thought processes that go into developing a solid storage area network.  If you would like to learn more, please contact Enterprise Networks Solutions to schedule a meeting.