The concept of Redundant Arrays of Independent (or Inexpensive) Disks (RAID) has been around for years. But, with technological increases in drive densities, drive speeds, and the amount of mission critical data being generated by new generation operating systems, and decreases in the cost of hard drives and RAID capable subsystems, RAID is becoming more useful and more prevalent to the enterprise. In the same way, a major application of RAID that is becoming more and more popular among larger enterprises with growing storage and access requirements is the Storage Area Network, or the SAN.
Storage Area Networks
Storage Area Networking is a subset of a group of technologies known as clustering, simply defined as the connection of two or more computing systems in such a way that they act as a single computer. Clustering is a popular strategy for implementing parallel processing applications because it maximizes the investment already made in computing systems. Clustering is becoming a critical capability and will be a key technology adoption for users wishing to satisfy their needs for high availability, scalability, and ease of management.
Storage and servers attached together via a high speed I/O interconnect, such as Fibre Channel, and managed by a software package is the basis for a SAN. It becomes a network where many computer systems can share storage resources and/or data. SANs go straight to the heart of the cost and performance issues that plague conventional networks, completely eliminating storage on the file server and its associated cost. And, since the SAN eliminates server data bottlenecks and dramatically reduces the use of network protocols to transfer data, performance is radically improved.
RAID with SANs
RAID is a key player in SANs. Since SANs are most useful for organizations with large, mission critical applications and online data storage requirements, the use of RAID systems becomes a necessity. RAID's redundancy, security, and reliability are vitally important to SAN environments.
In a standard setting, this system is adequate for the transmission of ordinary data (e.g., e-mail, word processing, operations) between clients and servers. The data requests in such an environment are small — averaging less than 200K per document. The servers can easily handle such requests, even to a fairly large number of clients. However, if the amount of data to be transferred is very large, like data stored in a RAID, a client-server network is likely to be inadequate. Examples of large data transfers include:
Storage Area Networks are designed to relieve the jams caused by the traditional client-server network. SANs permit the transfer of data without a server directly between servers, or imaging workstations and the storage using Fibre Channel — high-speed serial I/O — instead of the network stack. Clustered servers using SAN can scale past the 'client side' traffic jam by using distributed applications across multiple servers.
Data Storage Is Allocated Among Servers
The design of the SAN removes all of the bottlenecks associated with traditional client server networks — the server and the network stack — and replaces them with software and hardware which allow direct disk access via high speed Fibre Channel.
In addition to providing greater ease of disk management and improved server performance, SANs are highly efficient and reduce the total cost of disk administration by a factor of seven. Simple administrative clustering, which allows disks to be centralized via SAN technology, reduces the cost of managing storage from 55% to 15% of the cost of the hardware (IDC 1997).