Browse Prior Art Database

Distributed method for network bandwidth allocation

IP.com Disclosure Number: IPCOM000013570D
Original Publication Date: 2000-Dec-01
Included in the Prior Art Database: 2003-Jun-18

Publishing Venue

IBM

Abstract

Previous disclosures have delineated a design for bandwidth allocation in one shared resource such as the Ingress side or such as the Egress side of a network processor. The present disclosure concerns the distributed computational consequences of those designs through an entire switch or in multi-switch networks. Pipes are logical paths of aggregated flows. A pipe might be a DiffServ Behavior Aggregate Flow or an MPLS tunnel (or trunk). Each pipe might have a minimum guaranateed bandwidth value such as 1 Mbps. If the offered load in a pipe with such min is less than or equal to min, then the packets of that pipe should be expeditiously transmitted and not discarded at all. Each pipe might also have a maximum bandwidth limit such as 10 Mbps. If the offered load in a pipe with such a max is above its max for a significant time interval, then by use of discards the bandwidth in the pipe should be reduced to its max value. Three types of pipes might be present: 0 min max capacity (analogous to Expedited Forwarding (EF)