Brocade Communications Systems 53-0001575-01 Numbers above are labels for the links, see text

Page 11

SAN Design: March 29, 2001 3:18 pm

2.6HIGHLY AV AILA BLE SMALL FA BRIC CONFIGURA TION

FIGURE 6. HIghly Available Small Fabric

 

H

 

 

 

D

 

 

 

 

 

2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SW

 

 

SW

 

 

 

5

 

 

 

 

 

 

 

 

 

 

1

 

 

 

 

 

6

 

3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SW

 

 

SW

 

 

 

 

 

 

4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

D

H

 

 

Numbers above are labels for the links, see text

8port switches -- 20 N-port devices supported with max of 2 hops. Inter switch

Link failure always has alternate link available at hop count penalty

16 port switches -- 52 N-port connections

In this design, any switch can fail and there will still be an alternative path to the host and storage devices. [Assumes hosts/storage have intelligence to fail over.] Any single switch could be powered off for servicing and replaced in the fabric without loosing device connectivity.

This simple switch topology allows for multiple routes through the fabric ensuring that any single inter-switch link failure (GBIC or cable) will not result in loss of connectivity. BROCADE switches do automatic failover to alternate link and will recompute routing tables for all N-port devices.

The typical path through the fabric for a host requires only one hop to get to the storage. Should path 1 fail, traffic will route via 5 and 4 to get to disk. If the switch fails, the alternate HBA in the host can be used to still allow a single hop to the storage using path 6. Should paths 1 and 6 fail, there is a still a three hop path to the storage device (2-3-4). This provides for considerable resiliency and flexibility to fabric reconfigurations

There are a number of host and storage suppliers that can provide for detection of failed HBAs, failed paths, and failed ports on dual ported storage devices and that will use host based software to initiate a failover scenario. At this time failover is managed by hosts or by intelligent RAID systems that can detect the failure, it is not done by switch software

Storage in a fabric is globally available to all host elements (assuming a shared storage file system is in place). Where this is not the case, and storage is typically associated with a specific host, the storage is best placed on the same switch as the host to minimize inter-switch link traffic.

This design is appropriate when the fabric itself needs to be highly available; a single switch can fail and/or be taken off-line for maintenance and the fabric will still support all connected devices (devices do require one redundant entry point to the fabric)

53-0001575-01

BROCADE Technical Note

Page: 11 of 31

Image 11
Contents SAN Fabrics Design and Best Practices Guide Design and Best Practices Guide Fabric Topologies Tier Architecture Designs Single fabric, non-resilient Approach to take to SAN design for high availability Hosts Sample Switc H Configura Tions Two Fabrics -- Simplest Redundant Fabric Configuration Extended Fabric Example Numbers above are labels for the links, see text Design Guidance Edge Two Switch Core Star SAN Design Design Criteria /TESTED AN D Supported Configurations Star Topologies Possible with 2 switch cores Three Tier Design used in Fabric Aware Testing at Brocade MUlti Tier SAN Design across an Extended Link Twenty Switch Three Tier Design also a Star configuration FOS Fabric Bring UP Valid and Invalid Configurations using Inter Switch Links ZONING/NAME Space Switch HOP Coun T SAN MAN Agement NON Public Devic ES AN D Existin G FC-AL Disk Equipmen T Brocade Technical Note Glossary Gbic Topology Copyright