Browse Prior Art Database

HEARTBEAT MECHANISM FOR THE TIMED DETECTION OF TRANSMISSION CONTROL PROTOCOL CONNECTION FAILURES

IP.com Disclosure Number: IPCOM000249663D
Publication Date: 2017-Mar-15
Document File: 8 page(s) / 245K

Publishing Venue

The IP.com Prior Art Database

Related People

Gang Ding: AUTHOR [+3]

Abstract

A heartbeat mechanism is provided that operates in a layered architecture through the application of the standards-based Transmission Control Protocol (TCP) urgent data feature. The heartbeat mechanism allows for the accurate and precisely timed detection of TCP connection failures while resolving the deficiencies associated with current heartbeat approaches.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 27% of the total text.

Copyright 2017 Cisco Systems, Inc. 1

HEARTBEAT MECHANISM FOR THE TIMED DETECTION OF TRANSMISSION CONTROL PROTOCOL CONNECTION FAILURES

AUTHORS: Gang Ding

Vinnie Milano David McDonell

CISCO SYSTEMS, INC.

ABSTRACT

A heartbeat mechanism is provided that operates in a layered architecture through

the application of the standards-based Transmission Control Protocol (TCP) urgent data

feature. The heartbeat mechanism allows for the accurate and precisely timed detection of

TCP connection failures while resolving the deficiencies associated with current heartbeat

approaches.

DETAILED DESCRIPTION

State-of-the-art real-time failover capabilities may tolerate any single point of

software or hardware failure. In a distributed environment, a decisive factor for the failover

to work appropriately is the ability to detect network connection failures precisely within

a predefined stringent time window. For example, certain capabilities may detect outage of

a control path network connection (i.e., a server-to-server path connecting duplexed sides

of a critical processing node) within 500 milliseconds.

Some of these mission-critical applications use a heartbeat mechanism to monitor

connection state in order to notify the application when a connection is lost abnormally.

For example, it is abnormal for a connection to go down without first receiving a finish

(FIN) or reset (RST) packet over a Transmission Control Protocol (TCP) / Internet Protocol

(IP) connection. As described herein, the heartbeat mechanism is implemented via a

layered structure. More specifically, it is implemented in a product-specific

session/transport layer referred to herein as an external message transport (EMT), which

sits above the socket layer. The EMT layer hides the complexity of the contained socket

application program interface (API), and is optimized with various techniques for

Copyright 2017 Cisco Systems, Inc. 2

throughput improvement to render the required manipulation transparent to the

applications running above.

Two heartbeat mechanisms have been previously implemented in EMT: a User

Datagram Protocol (UDP) heartbeat mechanism and a TCP keep-alive heartbeat

mechanism. Both have limitations, as described below.

The UDP heartbeat mechanism works by sending heartbeats through a UDP

connection that is separate from the TCP connection (typically used for business data), and

the EMT binds the two connections together during connection establishment time.

However, it is impractical to send heartbeats through the same TCP connection used for

the business data because heartbeats are interleaved with business data in the same data

stream. As such, the receiver cannot parse the heartbeats out of the stream without reading

and buffering the business data intended for the upper layer (i.e. the business-logic layer).

As will be described below, the heartbeat mechanism provided herein addresses this

problem.

It is also impractical to send heartbeats through a separate TCP connection instea...