All networks generally include some form of storage. Traditional network storage was located inside, or directly connected to, individual file servers that were often scattered across workgroups throughout an organisation.
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
This resulted in a cumbersome, complicated, multiple-server environment that was virtually impossible to organise or secure. A storage area network (San) overcomes these problems by moving storage resources off of the common user network and reorganising those storage components into an independent, high-performance network. Storage performance is enhanced using a fast interface (e.g., 2 Gbps or 4 Gbps Fibre Channel) (FC) that connects storage servers and storage devices through an array of switches and hubs to form a fabric that supports both redundancy and high availability. San technology also supports important storage features, including disk mirroring, data backup/restoration, data archiving/retrieval and data migration.
San components and architecture
A storage area network is typically assembled using three principle components: cabling, host bus adapters (HBA) and switches. Cabling is the physical medium used to interconnect every San device. Sans can use both copper and optical fiber cabling, though the choice of medium depends on the speed and distance requirements of the San. Slower or shorter distance connections can be made through copper cables, while faster or longer distance connections are achieved through optical cables. Optical fiber cables can be single mode or multimode. Single-mode (or monomode) fiber is designed to carry only one light signal over long distances, while multimode fiber can carry multiple simultaneous light signals over short distances. Optical fiber also uses several different kinds of connectors, so it's important to select connectors that are compatible with other components of the fabric.
Each server or storage device in a San fabric requires an HBA. The HBA can exist as either an expansion card that fits into a compatible expansion slot in a server, or it may be a chip integrated directly into the server or storage device. An HBA typically offloads data storage and retrieval overhead from the local processor, improving the server's performance. Cabling is used to connect the HBA's port to a corresponding port on a switch.
A switch is used to handle and direct traffic between network devices. The switch accepts traffic, and then relays the traffic to the port where the intended destination device is attached. In a San, each storage server and storage device connects to a switch port. The switch then relays traffic to and from specific devices across the San -- this series of switched interconnections form the San "fabric," which can easily be scaled or changed. An intelligent switch serves the same basic functions but incorporates high-level San features like storage virtualisation, quality of service , remote mirroring, data sharing, protocol conversion and security.
San connectivity and protocols
Storage area networks are also defined by their interconnection scheme, which usually falls into either FC or iSCSI. FC technology is clearly the most popular approach for enterprise data center Sans. FC supports communication between servers and storage devices at 2 Gbps, though 4 Gbps implementations are now common, and 10 Gbps implementations are expected in the future. FC traditionally uses optical fiber cables to interconnect devices and is still employed over long distances. Today, short distance FC implementations can be achieved with coaxial and twisted-pair copper cables. FC can operate directly between two devices (point-to-point), or network multiple storage devices through a switch or arbitrated loop. FC technology is compatible with SCSI and IP protocols.
iSCSI is an emerging Internet Engineering Task Force standard that allows SCSI commands to support data storage and retrieval over Ethernet networks that include LANs, WANs and the Internet. By leveraging the broad acceptance of IP networks, iSCSI technology is expected to strengthen the San market and has already found acceptance in small and midsized organisations for basic San deployments. Since Ethernet networks generally work up to 1 Gbps, iSCSI isn't as fast as FC, which starts at 2 Gbps. However, iSCSI is less expensive than FC, and Ethernet is well-understood by any IT professional. In addition, 10 Gbps Ethernet is on the horizon and could also threaten FC's established position as the San networking technology of choice.
Creating a storage network is more involved than simply cabling servers and storage systems together. Storage resources must be configured, allocated, tested and maintained as new devices are added and enterprise storage requirements change. Management is a vital part of San operation, so it's important to select tools that can minimise the time and effort needed to keep a storage area network running.
Storage resource management (SRM) applications are designed to monitor and manage physical and logical San resources. Physical resources include storage arrays, RAID systems, tape libraries and FC switches, while logical storage features involve file systems and application-oriented storage elements (e.g. Oracle database files). It's usually best to select one tool that can provide centralised management of the entire storage infrastructure through a single console. Ideally, a centralised SRM tool should be able to detect storage resources, evaluate their capacity and configuration, and measure their performance. The SRM tool should also be able to affect changes to the configuration and support consistent policies across the various storage technologies being managed. San management tools are available from EMC Corp., Symantec Corp. (Veritas), McData Corp., Hewlett-Packard Co., IBM, Sun Microsystems Inc. and CA Inc.
In actual practice, selecting a San management/SRM tool can be an extremely challenging process -- usually because each tool accomplishes its suite of tasks in a unique way. Consequently, a good management tool should offer heterogeneous support, being able to accurately detect, discover and visualise a San across a variety of network equipment, storage systems and operating systems. The tool should provide meaningful monitoring and reporting features, including performance measurement, and that data should provide practical information that can help an administrator identify and resolve problems within the San.