Storage Area Network: Architecture

Storage Devices

Storage units are the heart of a SAN; these devices actually house the data. They are fully redundant and self-contained.

Storage Processors

A storage processor could be considered the brain of the system. They include one or two modules featuring a dedicated processor and memory for read and write caching. Among other things, they are responsible for RAID protection. Two of these modules are required for redundancy.

Storage Expansion

Link control cards (LCCs) allow the storage processor to be expanded. Multiple disk arrays can be added by simply daisy-chaining them off the storage processors.

Fibre Channel Hard Drives

Fibre Channel hard drives can spin at the rates of 10,000 or 15,000 RPM. Each hard drive is hot pluggable, meaning it can be added or removed while the system is powered on. The hard drives attach to a backplane that provides power and drive identification.

The storage processor communicates with the hard drives by an arbitrated loop running around the backplane. If two storage processors are installed, two loops will be running around the backplane providing redundancy.
If a break in a single (non-redundant) loop is detected, the drive will be moved to an offline status.

Automatic Rebuild

Storage systems achieve redundancy by having multiple sets of the same hardware. Data redundancy is achieved through RAID, tape backup, disk backups, or a combination of all three. In the event of data loss, provided a backup was made, the data can be restored. It will however take up precious time to restore the data. RAID provides an excellent first line of defense in data protection. A RAID array can be compromised if multiple hard drives in the RAID array fail. To protect against that, one or multiple hard drives can be designated a hot spare. Once configured, the hot spare will only be used in the event of a hard drive failure. If a failure occurs, data will automatically begin the rebuild process on the hot spare. Having a hot spare adds another level of redundancy to the hard drives themselves.

Automatic Failback

Automatic failback allows the data to be rebuilt back onto the originally failing hard drive. The hot spare then resumes its duties as a hot spare. For example, suppose five hard drives are configured in a RAID 5, and drive six is a hot spare. Drive three fails and the rebuild process begins on the hot spare. Once the failed drive three is replaced, the data is rebuilt onto drive three and the hot spare then resumes its watch for another failure.