Specifications
DATA CENTER BEST PRACTICES
SAN Design and Best Practices 58 of 84
•Error isolation and management: Most initiator errors are not propagated through to the fabric. Disconnecting
an upstream port, for example, does not cause a fabric rebuild. Most management activities on the Brocade
Access Gateway are also isolated from the fabric. One possible scenario is server administrators managing
the Access Gateways and storage administrators simply providing LUNs and zoning support for the servers
using NPIV.
•Increased resiliency: The Brocade Access Gateway supports F_Port Trunking, which increases the resiliency
of connections into the fabric. Losing a trunk member simply reduces the bandwidth of the upstream trunk.
While a few frames may be lost, no host connections are affected.
•Other: Hosts or HBAs can be congured to automatically fail over to another upstream link, should the one
they are using fail. The Brocade Access Gateway also implements many advanced features such as Adaptive
Networking services (a Brocade FOS feature), Trunking, hot code load, and Brocade Fabric Watch. Brocade
FOS v7.1 adds ClearLink D_Port support for Gen 5 Fibre Channel platforms, Credit Recovery, and Forward
Error Correction.
Constraints
The advantages of the Brocade Access Gateway are compelling, but there are constraints:
•Although benets are much more obvious for servers, the Brocade Access Gateway supports storage devices,
but the trafc must ow through the fabric, which has its own limitations.
•There are a maximum number of 254 NPIV connections per upstream port.
•The number of Brocade Access Gateways per switch is limited only by what the fabric switches can support.
The primary factors are:
••The total number of devices that attach to the fabric through the Access Gateways
•The number of devices per Access Gateway N_Port
•The total number of devices attached to the switch and fabric
See the Brocade Scalability Guidelines for details.
•The number of fabrics to which a single Brocade Access Gateway can be connected is limited to the number
of N_Ports on that Access Gateway. In general, most deployments require a single Access Gateway connection
to only one or two fabrics. Note that the ability to connect different upstream ports to different fabrics does
not reduce the requirement for redundancy. All attached servers should have dual paths to their storage
through different fabrics via separate Access Gateways.
Design Guidelines
Use the Brocade Access Gateway when you deploy bladed environments or have a lot of low-port-count switches
and when you need to connect different servers in different fabrics from a single-bladed enclosure. The Access
Gateway can be very valuable when you want to separate the management of blade enclosures so that the
enclosure is completely managed by server administrators, and the fabric is handled by storage administrators.
Management separation is provided through the NPIV connection, which allows the Access Gateway to be
managed separately by tools such as integrated blade server enclosure management tools without any adverse
effects on the fabric.
Monitoring
Monitoring is somewhat difcult for NPIV ows. Traditional SAN monitoring has been based at the port level
where hosts are connected. Multiple ows across ISLs and IFLs and into storage ports are common, but
multiple host behaviors into initiators are a relatively new concept. The Brocade Access Gateway has been
enhanced to include many features found in the standard version of Brocade FOS, such as Port Fencing, device
security policies, and Bottleneck Detection.