Browse Prior Art Database

Method for message passing by forced link fault on the stack links in the stackable switches

IP.com Disclosure Number: IPCOM000012355D
Publication Date: 2003-Apr-30
Document File: 5 page(s) / 57K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for message passing by forced link fault on the stack links in the stackable switches. Benefits include improved performance and improved ease of implementation.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 36% of the total text.

Method for message passing by forced link fault on the stack links in the stackable switches

Disclosed is a method for message passing by forced link fault on the stack links in the stackable switches. Benefits include improved performance and improved ease of implementation.

Background

        � � � � � The fixed port count of typical network switches necessitates stacking multiple switches together for scalability. In a stack, multiple switches are connected through high-speed links (referred to as the stack links)where as in the modular switches, multiple switch blades are connected through the stack links. A typical 2-switch stack has two switches, SW1 and SW2, connected together with, for example, two 1-GbE links (see Figure 1). A switch can force link fault at its end of the stack link usually by modifying a bit in the Media Access Control (MAC) register interface.

Fig. 1

        � � � � � With modular or stacked switches, the entire system must function as one large port count switch. Consequently, the address tables of all the switches (or the switch blades) must be synchronized periodically to account for network and/or address changes. Possible changes include:

•        � � � � Link (not the stack link) goes down and all the addresses residing on the corresponding port are deleted.

•        � � � � Address is aged out and deleted.

•        � � � � Stack link goes down and packets must be switched through alternative paths.

        � � � � � With managed switches, the management software running on the individual switches uses different protocols (RCP, for example) to exchange the information with other switches in the stack periodically. Various stacking algorithms use this information to update and address tables. As a result, the switches adapt to network changes. This approach results in several issues, including:

•        � � � � Protocols may require additional storage for maintaining various tables and are resources intensive.

•        � � � � Protocols require the capability to send and receive packets using the stack links.

•        � � � � Processor required should be powerful enough to perform the required tasks in addition to the typical tasks of address learning, address aging, and link state handling.

        � � � � � These issues can easily be solved with managed switches because they require processing capability for other functionality as well. However, with unmanaged or lightly managed switches (collectively referred to as unmanaged switches) where the intent is cost reduction, these issues are not easily solved. Typically, the unmanaged switches replace the management CPU subsystem with custom application–specific integrated circuit (ASIC) or field-programmable gate array (FPGA) devices. Additional memory can be very expensive and sending packets on the stack links may not be possible.

General description

        � � � � � The disclosed method enables message passing by forced link fault on the stack links in stackable switches. The method applies to any network switch/router that...