Browse Prior Art Database

Fast Transaction Processing Interface

IP.com Disclosure Number: IPCOM000119647D
Original Publication Date: 1991-Feb-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 3 page(s) / 108K

Publishing Venue

IBM

Related People

Chabanet, A: AUTHOR [+5]

Abstract

Enhancements described in this article are included in a new function of IBM Program called "X25 Network Control Program Packet Switching Interface" (NPSI): "Fast Transaction Performance Interface" (FTPI). The purpose of FTPI is to improve the performances of NPSI.

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

Fast Transaction Processing Interface

      Enhancements described in this article are included in a
new function of IBM Program called "X25 Network Control Program
Packet Switching Interface" (NPSI): "Fast Transaction Performance
Interface" (FTPI). The purpose of FTPI is to improve the performances
of NPSI.

      The transactions which will profit from FTPI, typically, credit
card or airlines companies reservations applications, are short in
terms of exchanged packets length.

      Today, this type of transaction is realized in the following
way:  a terminal or local cluster asks for the establishment of a
communication with a CTCP (Communication and Transmission Control
Program) running in a host server. The terminal or cluster
communicates with the NPSI function over an X25 network via a
switched logical link called a Virtual Circuit (VC).

      Upstream, the communications controller containing NPSI
Program, for instance, an IBM 3745 communication controller, is
attached to a SNA backbone. At the other end of the backbone, there
is a front end controller which is channel-attached to the host
server.

      In such an environment, the front end controller is overflowed
with too large a number of PIUs and this bottleneck may slow down all
local controllers performance and cause a response time degradation.

      Note: A controller slow down is mainly caused by too large a
number of PIUs, due to PIU time consuming process. In comparison, the
length of the PIUs has no significant impact.

      The FTPI solution consists of suppressing this bottleneck
thanks to a reduction of the number of PIUs crossing the front end
controller.

      FTPI Solution Description: Today, there is one LU-LU session
per X25 virtual circuit (VC) and the number of PIUs which cross the
SNA backbone is generally the same as the number of packets received
from or sent to the X25 network.

      Inbound Flow: To reduce the number of PIUs received from
network and destined for host CTCP, the packets received from the X25
network must be blocked within a given period of time (i.e., 100 ms)
in order to forward several packets in only one PIU. With current
session structure: one session per VC and according to the type of
aimed transactions, there will not be enough packets received on a
given LU-LU session to perform an efficient blocking within the
blocking time (i.e., 100 ms). Consequently, all the VC's
communicating with a given CTCP must be multiple...