Browse Prior Art Database

Management Station Protocol for Determining MAC-Layer Problems on a Heterogeneous Fiber Distributed Data Interface Ring

IP.com Disclosure Number: IPCOM000114040D
Original Publication Date: 1994-Nov-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 4 page(s) / 112K

Publishing Venue

IBM

Related People

Iliev, N: AUTHOR [+2]

Abstract

LAN error detection at the MAC layer will capture bit errors occurring on the link at the physical layer, malfunctioning receivers and transmitters, frame check sequencing errors, etc. In the Fiber Distributed Data Interface (FDDI) environment, the American National Standards Institute (ANSI) requires that MAC layer level error detection will be accomplished by providing a set of counters and a reporting process (*). The challenge, then, is to design a monitoring station protocol using only this information within a heterogeneous environment, for which it may not know the thresholds, etc., of some stations, and in which some forms of highly disruptive transient noise should be ignored. A decision as to whether or not a problem actually exists should be made in a timely manner, based upon finite data samples.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 38% of the total text.

Management Station Protocol for Determining MAC-Layer Problems on
a Heterogeneous Fiber Distributed Data Interface Ring

      LAN error detection at the MAC layer will capture bit errors
occurring on the link at the physical layer, malfunctioning receivers
and transmitters, frame check sequencing errors, etc.  In the Fiber
Distributed Data Interface (FDDI) environment, the American National
Standards Institute (ANSI) requires that MAC layer level error
detection will be accomplished by providing a set of counters and a
reporting process (*).  The challenge, then, is to design a
monitoring station protocol using only this information within a
heterogeneous environment, for which it may not know the thresholds,
etc., of some stations, and in which some forms of highly disruptive
transient noise should be ignored.  A decision as to whether or not a
problem actually exists should be made in a timely manner, based upon
finite data samples.  Ideally, this protocol should require a minimum
of processing time and memory storage.  Our protocol (the invention)
meets these challenges, while satisfying the above desirable
conditions.

      As dictated by the ANSI standard, three separate counters must
be maintained by each station on the FDDI ring for the purpose of
error detection:
 1.  Frame_Ct counts the number of frames passing by the station that
    are either valid (uncorrupted) or in error;
 2.  Error_Ct counts the number of frames passing by the station that
    are in error, and have not been reported to be in error by any
    other prior station (i.e., the E bit has not been set in the
    trailing delimiter);
 3.  Lost_Ct counts the number of frames passing by that station that
    are defined to be "lost," and have not been reported to be lost
    by any other prior station (i.e., an idle character has not been
    appended to end of a frame fragment).

      These counts are used to essentially construct the ratio of bad
frames to total frames (R) over a specified time interval (T) or
number of total frame arrivals (N), i.e., R(N) = % ( Delta %
'Error_Ct'  +  Delta % 'Lost_Ct' )/( Delta % 'Frame_Ct'  +  Delta %
'Lost_Ct' ) % app % E/N.  Delta indicates the change in these counter
values over T (or N).  If R(N) gt tau, where tau
 is a user- or designer-specified threshold, then a report is sent to
the ring manager station which which includes the raw counter values
at the end of the sample period, as well as R. Reports containing
this information continue to be sent at ANSI prescribed intervals (at
2, 4, 8, 16, 32, 32, ...  , 32 sec intervals).  If R(N) drops below
tau during this process, a new report is sent, reinitializing the
reporting interval sequence.  Let E sub 0 = Error_Ct + Lost_Ct and N
sub 0 = Frame_Ct + Lost_Ct from the initial report; similarly, let E
sub k and N sub k be the same counter value sums from the kth report
following the initial one.  The managing station may...