FIBRE CHANNEL and IP SAN INTEGRATION(2)

The maturity and mission-critical deployment of Fibre Channel (FC) in storage area networks (SANs) creates a unique class of multi-terabit networks with demanding throughput, latency, scalability, robustness, and availability requirements. This paper revie

FC. It also examines several protocol and design considerations, system-level behaviors, and areas that need further research and enhancement. This paper leverages the efforts of many engineers, architects, and researchers from the industry. The paper uses their findings and recommendations, and tries to relate them to SAN applications. 2.0 The FC SAN Today2.1 FC SAN OverviewFC technology [1] and product deployment has evolved from 1 gigabit per second (Gbps) to 2 Gbps links, and there is development to introduce 4 Gbps and 10 Gbps links. An FC network or fabric is a multi-terabit, low-latency switching network, mainly used to interconnect servers to storage. Although a FC fabric is designed to support any-to-any connectivity, the actual use tends to be some-to-some. Each server talks to a few storage devices or each storage device talks to a few servers, with occasional traffic for backup or other purposes involving devices shared by many sets of storage and servers. Deployment of mid-range to high-end FC fabrics is based on FC directors [2], which are highavailability switches with high-aggregate switching bandwidth and high port density. For the edge part of a large or small fabric, smaller and lower-cost FC switches are typically used. Directors and switches use one or more interswitch links (ISLs) to connect and form a larger fabric. It is common to deploy one or more isolated FC fabrics, called SAN islands. SANs are also extended to campus, metropolitan, and wide-area distances using T1/T3, ATM, IP, SONET, dark fiber, and DWDM technologies.HostServers/Blade ServersSwitch Switch Switch Switch SwitchSwitchSwitchSwitchSwitchSwitchStorageDirector DirectorDirector DirectorStorageStorageStorageInter-Switch Link (ISL)SwitchSwitchSwitchSwitchSwitchSwitchSwitchSwitchSwitchSwitchServers/Blade Servers Figure 1 Example of a Large FC Fabric Figure 1 shows an example of a large (approximately 1000 node) fabric, with directors and switches configured to provide high availability and high-aggregate bandwidth. Servers are typically aggregated at the edge of the fabric, and storage arrays are typically configured near the core of the fabric. It is typical to over-subscribe server link bandwidth in comparison to storage array link bandwidth (more servers with respect to a given storage array). A network of directors forms the core (or backbone) of the fabric.102