Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

METHOD FOR HANDLING MULTIPLE SESSION TIMEOUTS

IP.com Disclosure Number: IPCOM000009331D
Original Publication Date: 1999-Jun-01
Included in the Prior Art Database: 2002-Aug-16
Document File: 3 page(s) / 124K

Publishing Venue

Motorola

Related People

Tolya Agulnik: AUTHOR [+2]

Abstract

Consider a high performance task managing simultaneous sequential packet data communica- tions sessions with multiple remote network devices. Such a task requires a method for detecting communication failures with each device indepen- dently. Typically, the detection method involves a session timeout while waiting for a response from one of the devices in which a session is ongoing. The timeout period is generally much longer than the response time. Since, in general, the communi- cation sessions with the devices are not synchro- nized with respect to all devices, the session time- outs must be managed independently with respect to each ongoing session.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 3

MOTOROLA Technical Developments

  METHOD FOR HANDLING MULTIPLE SESSION TIMEOUTS

by Tolya Agulnik and Gary Soibel

BACKGROUND

  Consider a high performance task managing simultaneous sequential packet data communica- tions sessions with multiple remote network devices. Such a task requires a method for detecting communication failures with each device indepen- dently. Typically, the detection method involves a session timeout while waiting for a response from one of the devices in which a session is ongoing. The timeout period is generally much longer than the response time. Since, in general, the communi- cation sessions with the devices are not synchro- nized with respect to all devices, the session time- outs must be managed independently with respect to each ongoing session.

PRIOR ART

  Each session communicating with a separate device spawns a timer of length TIMEOUT each time it sends a packet to the device. A communica- tion timeout error is indicated if a response is not received from the device prior to timer expiration.

PROBLEM DESCRIPTION

  Prior art requires management of one timer per session per packet sent. This may cause degradation of performance since every time the response packet is received the timer must be restarted. In addition, under normal operation, communication timeout errors would not occur, and there is no need to

unduly stress the task. The number of timer start/restarts required to complete all sessions is given by

(l)N'=p*s where p is the number of packets per session s is the number of sessions

SOLUTION

  The main idea is to have only one timer running for as long period of time as possible.

  Initialization - When the first session is initialized, start a timer of length equal to the TIMEOUT period

  While Sessions are in Progress - Update a session-based time stamp each time a packet is sent for the corresponding session.

  When the timer expires, using the recorded ses- sion time stamps, check each session to see if a communication error has occurred for that session. Then using the earliest session timestamp t, start a timer which expires at t + TIMEOUT.

  Considering the diagram, on next page, we can see the advantage of using this technique to manage multiple sessions with independent timeouts. (For graphical purposes, assume that there is no delay between receiving a response packet and sending out the next packet).

0 tdomm,a, Inc. ,999 215 Jum? 1999

[This page contains 15 pictures or other non-text objects]

Page 2 of 3

Developments Tqchnical 0 M MOTOROLA

timerx+l...