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

OPTIMIZED AGING MECHANISMS FOR MAC ENTRIES

IP.com Disclosure Number: IPCOM000243580D
Publication Date: 2015-Oct-02

Publishing Venue

The IP.com Prior Art Database

Abstract

An optimized MAC aging mechanism is described which reduces a significant amount of overhead associated with locks to correspondingly increase performance. The mechanism uses a Least Recently Used (LRU) list separate from a MAC table, and the LRU list is updated every time a MAC is added or refreshed. For optimization, just one lock is proposed, but with two modes - READ and WRITE. This lock is used to protect data that allows multiple readers in parallel, but only one writer. If there is continuous traffic from one source, as is typical, a forwarding path will end up burning a significant amount of cycle’s updating the LRU list. The optimizations proposed herein avoid this at the cost of MAC aging accuracy. With LRU optimizations, the tradeoff is between better throughputs versus accurate MAC aging.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 18% of the total text.

Page 01 of 14

OPTIMIZED AGING MECHANISMS FOR MAC ENTRIES

ABSTRACT


[0001]An optimized MAC aging mechanism is described which reduces a significant amount of overhead associated with locks to correspondingly increase performance. The mechanism uses a Least Recently Used (LRU) list separate from a MAC table, and the LRU list is updated every time a MAC is added or refreshed. For optimization, just one lock is proposed, but with two modes - READ and WRITE. This lock is used to protect data that allows multiple readers in parallel, but only one writer. If there is continuous traffic from one source, as is typical, a forwarding path will end up burning a significant amount of cycle's updating the LRU list. The optimizations proposed herein avoid this at the cost of MAC aging accuracy. With LRU optimizations, the tradeoff is between better throughputs versus accurate MAC aging.

BACKGROUND


[0002]In Ethernet and other learning-based forwarding techniques, aging mechanisms are required to remove old entries, such as Media Access Control (MAC) entries. Specifically, switches use aging to keep Ethernet switching tables current. For each MAC address in the Ethernet switching table, the switch records a timestamp of when the information about the network node was learned. Each time the switch detects traffic from a MAC address that is in its Ethernet switching table, it can update the timestamp of that MAC address. Conventionally, a timer on the switch periodically checks the timestamp, and if the MAC address of a node is older than the value set, the switch removes that MAC address from the Ethernet switching table. This aging process ensures that the switch tracks only active MAC addresses on the network and that it is able to flush out from the Ethernet switching table MAC addresses that are no longer available.


[0003]With the increase in switch size, MAC aging is a complex task. With multiple ports, switches use a read/write lock scheme to implement MAC aging, and taking and releasing locks is an expensive operation. For example, to execute a simple instruction with lock takes about four times more compared to executing the same instruction without locks.

BRIEF DESCRIPTION OF THE DRAWINGS


[0004]The present disclosure is illustrated and described herein with reference to the various drawings, and in which:

DETAILED DESCRIPTION


[0005]In various exemplary embodiments, an optimized MAC aging mechanism is described which reduces a significant amount of overhead associated with locks to correspondingly increase performance. The mechanism uses a Least Recently Used (LRU) list separate from a MAC table, and the LRU list is updated every time a MAC is added or refreshed. For optimization, just one lock is proposed, but with two modes - READ and WRITE. This lock is used to protect data that allows multiple readers in parallel, but only one writer. If there is continuous traffic from one source, as is typical, a forwarding path will end up burning a signific...