Browse Prior Art Database

Fibre Channel - Low-Cost Redundant Loop

IP.com Disclosure Number: IPCOM000112061D
Original Publication Date: 1994-Apr-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 71K

Publishing Venue

IBM

Related People

Truestedt, HL: AUTHOR

Abstract

A loop topology provides a low-cost connectivity to a number of devices. However, a loop is a single point of failure which affects the entire group of devices during a failure. Described is a method to provide redundancy (and better performance) and still remain low-cost.

This text was extracted from an ASCII text file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 52% of the total text.

Fibre Channel - Low-Cost Redundant

Loop

      A loop topology provides a low-cost connectivity to a number of
devices.  However, a loop is a single point of failure which affects
the entire group of devices during a failure.  Described is a method
to provide redundancy (and better performance) and still remain
low-cost.

      The ANSI Fibre Channel Standard is using an Arbitrated Loop
(i.e., before sending any frames, the port must arbitrate (obtain
ownership) of the loop).  Some knowledge for the ANSI Fibre Channel
Standard (FCS) and FC-PH REV 2.2, January 24, 1992 is assumed for the
remainder of this article.

      Fig. 1 describes the loop model with four N_Ports (A, B, C, and
D).  Fig. 2 is a high-level view of an LN_Port (N_Port with Loop
capabilities provided by the Loop Port Logic (LPL)).

      The ANSI Fibre Channel Arbitrated Loop uses one serial loop to
provide a low-cost interconnection.  If this single loop fails, the
entire loop (and all of its nodes/devices) is unavailable.  In
addition, the Arbitrated Loop may only be used by one arbitrator and
one destination at a time, thereby reducing accessibility and
performance of other nodes/devices which may be trying to get access
to the loop.

      One solution is to have dual ports on each device (i.e.,
duplicate Fig. 2).  This will provide redundancy and multiple device
access (since there is more than one loop available).  However, the
N_Port logic is expensive.  Adding more logic to manage which N_Port
sends/receives frames would add even more cost.

      The following model provides better robustness and performance
than the single loop without the added cost of duplicating ports.
The only additional cost of the Redundant Loop is in the extra
cables, connectors, and driver/receivers.  Fig. 2 is modified as
depicted in Fig...