FW 05.01.00 and SW 07.01.00 HP StorageWorks SAN High Availability Planning Guide (AA-RS2DC-TE, June 2003)

Planning Considerations for Fibre Channel Topologies
96 SAN High Availability Planning Guide
The simplest core-to edge fabric has two or more core switching elements that
may or may not be connected (simple or complex). In a simple core topology, as
shown in Figure 39 on page 94, core switches are not connected. In a complex
core topology, as shown in Figure 40 on page 95, core switches are connected.
The figure also illustrates a topology where the core is a full-mesh fabric.
Each edge switch connects (through at least one ISL) to each core switch, but not
to other edge switches. There are typically more device connections to an edge
switch than ISL connections; therefore, edge switches act as consolidation points
for servers and storage devices. The ratio of ISLs to device connections for each
switch is a function of device performance. For additional information, refer to
“ISL Oversubscription” on page 99.
Fibre channel devices (servers and storage devices) connect to core or edge fabric
elements in tiers. These tiers are defined as follows:
■ Tier 1 — A Tier 1 device connects directly to a core director or switch. Tier 1
devices are typically high-use or high-I/O devices that consume substantial
bandwidth and should not be connected through an ISL. In addition, IBM
fiber connection (FICON) devices cannot communicate through E_Ports
(ISLs) and must use Tier 1 connectivity. For additional information, refer to
“FCP and FICON in a Single Fabric” on page 108.
■ Tier 2 — A Tier 2 device connects to an edge switch and Fibre Channel
traffic from the device must traverse only one ISL (hop) to reach a device
attached to a core director or switch.
■ Tier 3 — A Tier 3 device connects to an edge switch and Fibre Channel
traffic from the device can traverse two ISLs (hops) to reach a device attached
to a core director or switch.
Fabric Island
A fabric island topology connects several geographically diverse Fibre Channel
fabrics. These fabrics may also comprise different topologies (cascaded, ring,
mesh, or core-to-edge), but may require connectivity for shared data access,
resource consolidation, data backup, remote mirroring, or disaster recovery.
When connecting multiple fabrics, data traffic patterns and fabric performance
requirements must be well known. Fabric island connectivity must adhere to
topology limits, including maximum number of fabric elements and ISL hop
count. It is also essential to maintain data locality within fabric islands as much as
possible, and to closely monitor bandwidth usage between the fabric islands.