There are many reasons why you may want to buy a SAN. Some benefits of a SAN include the following:
- A SAN allows you to pool disk storage resources and allocate to systems only as needed instead of allocating entire disks to systems.
- With a SAN, storage is shared at the block level instead of at the file or directory level. This means that individual systems can format the disk space as needed or use raw volumes.
- Some applications such as Exchange 2010 are only supported on block-level storage.
- Companies may want to use diskless systems (such as blades) to boot from the SAN.
Storage area network basics: What is a SAN?
The first storage area network basic concept you need to know is the general definition of a SAN. A SAN is shared block-level storage made available through a dedicated network. This is in contrast with network-attached storage (NAS), which is file-level storage accessed via a regular IP network. It is often confusing because both are accessed via a network -- the important difference is that a SAN is exclusive to block-level storage. This means that disk storage is seen by a server as if it were local and not on a network. Also worth remembering is that a storage array by itself is not a SAN; a SAN is the combination of the shared storage array, the network and other devices as a whole connecting it to the servers or hosts.
Fibre Channel vs. iSCSI SANs
There are two main types of SANs available: Fibre Channel (FC) and iSCSI, which are essentially protocols used to access block-level storage. Early SAN offerings focused on FC. When FC SANs were introduced, they mostly appealed to enterprise-class IT organization because of its higher cost -- an FC network requires special adapters called host bus adapters (HBAs), FC cables and FC switches. iSCSI came after Fibre Channel and offers a much more affordable alternative to FC because it can leverage a regular IP network to connect the servers to the storage array.
iSCSI SANs still require special adapters, or initiators, to convert SCSI protocol to IP but do not require special network components or cabling. In fact, it’s possible to use a regular network card (NIC) in conjunction with an iSCSI software initiator to implement an iSCSI SAN without special equipment. The only thing you would need to do this is a shared storage array that has iSCSI ports.
What works for SMBs?
Generally speaking, iSCSI SANs are more accessible to small- to medium-sized businesses (SMBs) than Fibre Channel SAN solutions. The main reason is the cost of equipment. There is also the question of acquiring new skills to manage the Fibre Channel network when a FC switch is used. Because of the difficulties FC SANs can present to small businesses, iSCSI SANs are more appealing to SMBs. Besides the requirement of a storage array and special adapters or initiator software, you can deploy an iSCSI SAN using a conventional high bandwidth IP network while leveraging existing IP network skills.
Buying your first SAN: A checklist
Follow this high-level list of components your SMB would need to deploy an iSCSI SAN:
- A shared storage array with iSCSI ports.
- Storage management software. This will help you to create and assign storage shares. Keep in mind that storage management software may or may not come with the storage array.
- High-bandwidth dedicated IP network segment. While iSCSI can run on a shared network with other IP traffic, a dedicated network segment is recommended to avoid performance degradation.
- iSCSI hardware and software initiators (host adapters) for the servers that will share storage on the SAN.
Other SAN implementation considerations
When considering the implementation of a SAN, there are a few other storage area network basics that must be considered to ensure the solution will meet the objectives and requirements. Some of these elements include:
- Number of hosts supported. The storage array must be able to support the number of servers that will share storage. This is dictated by the number of LUNs that can be created. For example, an entry-level device that only supports 14 LUNs will limit the number of servers that can share storage to 14.
- Redundancy. Moving your data to centralized shared storage also means the risk of losing access to all data in the event of a storage array failure. Selection criteria must therefore include internal components redundancy and the ability to replace them without taking an outage (hot swappable).
- Data protection. Very much aligned with the previous bullet, data backup and data protection services must also be considered. While the solutions mentioned earlier also come with data protection features such as snapshot or point-in-time copy capabilities, a proper offsite data protection scheme is still required. These arrays support local or remote replication, but from an SMB perspective, the cost of a second storage array can be prohibitive and hard to justify. In such case, a traditional data backup (i.e., tape backup) might still be required for offsite data protection.
As with any other IT solution, developing an understanding of storage area network basics should come first, followed by research on products and features.
About this author: Pierre Dorion is the data center practice director and a senior consultant with Long View Systems Inc. in Phoenix, Ariz., specializing in the areas of business continuity and DR planning services and corporate data protection.
This was first published in March 2011