Than Current Storage Options?">
In order to truly appreciate the benefits of a SAN we need to carefully examine the current state of distributed systems and their storage devices.
In traditional server-attached storage, all access to that servers storage resources are made through that single computer. As a result, the server itself can become a bottleneck if it needs to run too many applications or support too many users at once. Plus, users need a one-to-one relationship with a computer to use its resources. If only one computer on a network is running a particular application such as an e-mail system or a database, users have to connect with that machine.
Furthermore, current server-attached storage lacks the flexibility to move storage resources. For example if a large UNIX server has a 100 GB of spare storage space, an IT manager cannot reallocate this resource to a storage-starved NT or Novell server. In a SAN environment, IT managers (using storage management tools) have the ability to resize the partitions on their centralized storage repositories and reallocate unused storage space to needy servers.
"SANs make it easy to add storage without the necessity of taking down a server. Since a SAN is a network, it is independent of any single device connected to it."
SANs make it easy to add storage without the necessity of taking down a server. Since a SAN is a network, it is independent of any single device connected to it. To add a device, all thats necessary is to give the device a LUN (logical unit number) and connect it to the SAN. Once storage is added to a SAN, an IT manager can use this newly added storage to increase the size of a given storage volume. This makes the newly added storage available to all users and applications without having to reprogram every workstation or application. This is a key feature that makes SANs useful to large enterprises that are experiencing rapid growth.
Without a SAN, adding storage to a server requires taking that server down and interrupting operations, to back up data to tape. After data has been backed up, an administrator would have to blow away their current RAID setup, add the new disks to the enclosure, and restore their data back to the rebuilt storage unit. While this operation was acceptable for many businesses in the past, now that we are in the age of 24/7 E-Commerce, unnecessary downtime has become intolerable.
Partitioning is a method by which pooled storage is subdivided into smaller, more manageable volumes that can be allocated to specific applications, servers, clusters, etc. For example, a RAID system first pools multiple physical hard drives into a single large virtual volume. This large virtual volume can then be partitioned into smaller volumes. Each of the partitioned volumes on a RAID subsystem can be the same size as a single hard disk, but the volumes benefit from the redundancies and performance advantages inherent in RAID technology. Partitioning lets operating system file systems use disk space more efficiently. It gives system administrators flexibility when planning security and assigning users their storage access.
SANs can be partitioned just like a RAID system, only the partitioning takes place over the entire storage network rather than individual storage subsystems. So, if a SAN is composed of multiple RAID systems of various speeds, plus tape and optical archives, each partition of the SAN may include portions of these different storage resources.
Each partition is controlled by its own server, server cluster, or shared file system. System administrators have the freedom of assigning storage resources to different partitions in any combination. Each department in an organization might be allocated its own partition, and can determine its own file migration, redundancy, and backup requirements.