Enterprise data centers often incorporate large director-class switches that offer massive scalability for the storage area network, yet can be managed as a single platform. In addition, director switches feature fault tolerance, nondisruptive software upgrades, hot-swappable components, redundant power and cooling systems, and advanced security features. Director switches can frequently support multiple network protocols and can even be logically organized into multiple virtual switches. Heterogeneous server and storage system support is critical for a director switch. The two principle issues with director switches are cost and management: Directors are significantly more expensive to acquire than other switch classes, and the general lack of friendly wizard-driven interfaces usually means that management tasks can be more cumbersome and time consuming.
You'll typically find products from Cisco Systems and Brocade Communications Systems in the director space. For example, Cisco's MDS 9513 is one of the largest director switches available, sporting as many as 528 Fibre Channel ports in a single chassis, each port capable of 1 Gbps, 2 Gbps, 4 Gbps or 10 Gbps. The MDS 9513 includes support for VSANs and VSAN routing, along with advanced security features. Brocade's DCX Backbone director supports up to three hundred and eight four 4 Gbps or 8 Gbps ports in one chassis, and connects to other Brocade directors, such as the 48000 and Mi10K.
QLogic offers a SANbox 9000 director, which at 128 ports is smaller and less expensive than the Brocade and Cisco directors, but it lacks the mainframe connectivity and redundancy of its rivals' products.
Along with large modular, redundant and chassis-based directors, organizations also use smaller fixed-port fabric switches in departments and outside the data center. The Brocade 5000 is 32-port switch, Cisco's MDS 9124 is a 24-port switch and the QLogic SANbox 5802 is a 24-port switch that supports 8 Gbps connectivity. Besides being smaller than directors, Fibre Channel switches lack features, such as automatic failover and hot swappable components that would allow them to continue to work in event of a failure. Fabric switches are also known as edge switches.
In terms of SAN switches, "intelligence" suggests the presence of advanced features, such as storage virtualization, remote mirroring, data sharing, protocol conversion, quality of service (QoS) management and strong authentication/security features. As storage networks become more sophisticated and support a growing range of storage products, adding intelligence to the switch offers compelling benefits -- most notable is the centralization of network-based services. Rather than managing network services handled by a variety of applications running through individual servers in different locations around the network, managing an intelligent switch potentially lowers labor costs and software licensing costs (e.g., volume managers) and improves resource utilization by reducing overprovisioning.
However, intelligence has nothing to do with the size of a switch. Thus, director-class and intelligent switches are not the same thing. A director switch may not have many intelligent features, if any, and often must be configured with service modules to support intelligent functions. Conversely, intelligent features are appearing more frequently in blade switches and even in some lower end products, like edge/workgroup switches.
Brocade, Cisco and QLogic sell intelligent switches.
Blade switches are basically modular switches designed to fit in blade chassis produced by major manufacturers like Dell, IBM, and Hewlett-Packard. Although there are no features or capabilities unique to blade switches, the blade form factor offers a substantial advantage to blade server users. By mounting a blade switch, other blade servers in the chassis can be connected through the blade switch to the rest of the network. This is cleaner and more convenient than cabling each blade server to an external switch.
However, it's important to remember that blade switches generally offer limited scalability, and they are more expensive to deploy because blade chassis are not yet mainstream. Blade servers may also experience issues booting from the network through a blade switch.