Embedded switching adds reliability

Embedded storage switching technology makes it feasible for storage systems to incorporate 2G bit/sec switched network connectivity within a storage array. Some benefits of embedded storage switching include much higher reliability, better performance and the ability to add drives without degrading performance.

Shared bus architectures are used in the back end of many storage systems, rendering each disk or tape drive within the storage array a single point of failure. This dramatically increases the risk of an entire disk array going offline as the result of a problem with a single drive.

Whether the front end of a storage-area network or network-attached storage system uses Fibre Channel, iSCSI or IP, the storage system controller translates requests from the front end to the back end using the raw Fibre Channel Arbitrated Loop (FC-AL) shared bus protocol.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

Embedded storage switching technology makes it feasible for storage systems to incorporate 2G bit/sec switched network connectivity within a storage array. Some benefits of embedded storage switching include much higher reliability, better performance and the ability to add drives without degrading performance.

Shared bus architectures are used in the back end of many storage systems, rendering each disk or tape drive within the storage array a single point of failure. This dramatically increases the risk of an entire disk array going offline as the result of a problem with a single drive.

Whether the front end of a storage-area network or network-attached storage system uses Fibre Channel, iSCSI or IP, the storage system controller translates requests from the front end to the back end using the raw Fibre Channel Arbitrated Loop (FC-AL) shared bus protocol.

A new architecture

Arbitration and data flow of legacy FC-AL loops must progress through all devices in the loop. Each device along the path adds latency - and more importantly, an additional failure point that reduces reliability.

Legacy FC-AL loops operate through the mechanism of the controller, first arbitrating for control of the loop and then sending a command to a drive to prepare for data to be written to the drive or to request data from a specific location.

Back-end switching, however, uses the FC-AL protocol in a switched fashion, bringing point-to-point connectivity to each individual drive.

Embedded storage switching combines the crucial elements of complete integration of a crossbar switch core, embedded serializer/deserializers, 1- and 2-gigabaud bandwidth capabilities, and diagnostics to a single switch on a chip.

Storage systems physically arrange several disks into a single enclosure known as a Just a Bunch Of Disks (JBOD). Before the availability of the new level of integration brought by embedded storage switches, placing switching within a JBOD was impractical in modular storage systems because of real estate, power, heat and pricing issues. When embedded storage switching is added to a JBOD it becomes a Switched Bunch of Disks (SBOD).

Now when an initiator arbitrates for control of the system, instead of the arbitration proceeding through all the devices, such as in a shared bus architecture, the arbitration goes only to the switch matrix and back to the initiator. The process is very fast and greatly reduces system latency. After the initiator gains control, it can open the target drive and begin communications. Data packets sent between the initiator and drives are now point to point, and multiple conversations can take place at the same time.

Whether an embedded back-end switch provides connectivity from a controller to several JBODs or all the way to the individual hard disk drive using SBODs, embedded switching enables continuous reliability, availability and serviceability.

Embedded switching gives IT managers tools to deploy automatic maintenance, monitoring and repair. Each embedded storage switch port retimes the low-level signal, increasing signal integrity and reducing system jitter. Back-end embedded storage switches are in the ideal position to monitor traffic and signals, and to diagnose problems.