Fabric switches can support dedicated N_Port connections for full bandwidth to a single node or can support shared NL_Port connections for arbitrated loop nodes. The most common loop devices on a fabric switch are JBODs, which appear on a switch port as a series of downstream NL_Ports on a common loop. It may also be reasonable, however, to group servers on a loop segment if the bandwidth required per server is well below the 1Gbps or 2Gbps bandwidth provided by the switch. The combination of loop and dedicated fabric nodes on a switched infrastructure provides flexibility in SAN design and enables cost-effective use of switch ports. As shown in Figure 4-17, the customer can use an arbitrated loop hub to allocate shared bandwidth for servers that have lower performance requirements while devoting full bandwidth to high-performance servers. JBODs do not have the throughput of RAID, but they offer an economical alternative to high-performance RAID storage. Figure 4-17. Fabric switches, loop hub, and loop JBODs on a single Fibre Channel SANThe ability to integrate loop and fabric components in a single solution is also useful for customers who have already deployed loop-based SANs in various departments and now wish to tie them together into an enterprise-wide storage network. Although switch vendors are always willing to sell more switches in each department, leveraging the investment the customer has already made lowers the deployment costs and still accommodates assignment of bandwidth as required by the storage application. |