Browse Prior Art Database

Dynamic End Node CP-CP Session Switching Using Alternate Network Node Servers

IP.com Disclosure Number: IPCOM000122106D
Original Publication Date: 1991-Nov-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 4 page(s) / 176K

Publishing Venue

IBM

Related People

Cossack, MA: AUTHOR [+3]

Abstract

Described is a technique for allowing an Advanced Peer to Peer Networking (APPN) end node to reestablish control point sessions with an alternate network node server following a control point session outage with its original server. This technique provides the function dynamically without operator intervention and without disrupting any communication links that have already been established by the end node. Other methods of performing this function required a link-level exchange identification (XID) to convey the need to establish a new server.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 37% of the total text.

Dynamic End Node CP-CP Session Switching Using Alternate Network
Node Servers

      Described is a technique for allowing an Advanced Peer to
Peer Networking (APPN) end node to reestablish control point sessions
with an alternate network node server following a control point
session outage with its original server.  This technique provides the
function dynamically without operator intervention and without
disrupting any communication links that have already been established
by the end node.  Other methods of performing this function required
a link-level exchange identification (XID) to convey the need to
establish a new server.

      This invention solves both of the problems regarding control
point to control point (CP-CP) session reestablishment following a
CP-CP session outage with an original network node server.  It
provides the function without operator intervention and it eliminates
the disruption of active user sessions to carry out the task.

      The reason that XID exchanges were required to reestablish
CP-CP sessions with an alternate server is because the APPN
architecture stated that an end node could only request one network
node to be its server (during XID exchange) at any given time.  Once
that relationship was established, the end node would establish its
CP-CP session with its network node server, and the network node
would also establish its control point session with the end node
(since a logical CP-CP session between two nodes actually involves
two separate sessions - with each node being the 'owner' of its
session).  The other network nodes would be told by the end node
(during XID exchange) that network services were not required.  If
the CP-CP session ended with the original network node server, the
end node was unable to reestablish the CP-CP session with an
alternate server (because it had told the alternate server that it
was already being served during their previous XID exchange). Refer
to the figure.  If Portland has requested network node services from
Seattle, then Houston and Chicago will be told that Portland does not
require services when their links with Portland are established.  If
the CP-CP session is lost between Portland and Seattle, then Portland
would need to deactivate and then activate the link to either Houston
or Chicago in order to join the APPN network again.

      The solution involves changing two of the premises stated
above:
1.   An end node only requests one network node be its
     server at a time.
2.   Having the network node establish its CP-CP session
     immediately following the XID exchange with the end
     node requesting its services.

      As an end node supporting this new function, it will indicate
during its XID exchanges that it is requesting services from every
adjacent node (provided the adjacent node is in a configured list of
potential network node servers).  The first network node in the
server list that the end nod...