Browse Prior Art Database

Alive Protocol

IP.com Disclosure Number: IPCOM000111971D
Original Publication Date: 1994-Apr-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 97K

Publishing Venue

IBM

Related People

Lozinski, ZA: AUTHOR

Abstract

Disclosed is a technique for two distributed processors to mutually detect failures in each other. In an embodiment it is applied to a PS/2 system unit and ISDN co-processor adapter to prevent excessive communication line changes if failure occurs such that recovery action cannot be invoked. It is applicable to workstations which use co-processor adapters, especially ISDN communication adapters.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Alive Protocol

      Disclosed is a technique for two distributed processors to
mutually detect failures in each other.  In an embodiment it is
applied to a PS/2 system unit and ISDN co-processor adapter to
prevent excessive communication line changes if failure occurs such
that recovery action cannot be invoked.  It is applicable to
workstations which use co-processor adapters, especially ISDN
communication adapters.

      An ISDN adapter card for a PC contains PC software and adapter
microcode which has the function to set up and clear calls across the
ISDN.  A part of the adapter software runs in the PC under OS/2 and
another part runs on the ISDN Co-processor/2.  If an error occurs
either in the PC software component, or in the adapter microcode,
then it may not be possible to clear any connections until the PC is
rebooted, or powered off.  ISDN connections are usually charged in
the same way as telephone calls, by the length of time the line is
connected.  In the case of an internal error, the user may be paying
for an unusable connection, possibly for several hours.  This
combination of an adapter card and supporting PC software is common
to most ISDN solutions  for the PC and the problem is common to many
ISDN adapters.

      The component within the PC software that is responsible for
establishing and clearing calls is called the Port Connection Manager
(PCM) and is further divided into two subcomponents.

      Port Connection Manager (Workstation) runs on the PC.  PCM-W is
responsible for accepting application requests to setup and clear
calls, which are passed to it from OS/2 which it  sends to the
adapter.  Finally, responses from the adapter are returned to OS/2.

      Port Connection Manager (Adapter) runs on the adapter.  PCM-A
is responsible for processing requests from the workstation and
signalling these requests to the ISDN, using the standard ISDN call
control protocol.  It is also responsible for receiving messages from
the ISDN, for example the notification of an incoming call, and
communicating these up to the workstation.

      One potential problem with the above split of the Port
Connection Manager software into two subcomponents is what happens if
one of the two parts fails.  As both subcomponents run as processes
on their respective processors, the particular failure to be detected
is when either of the processes hangs, or is abnormally terminated.
If this failure is unrecognised, then any connected B channels will
remain connected, and the customer will be charged until the machine
is re-IPLed.  It is necessary to detect if one of the Port Connection
Manager components has failed, and to define what actions must tak...