Abstract

U4EA Technologies commissioned The Tolly Group to evaluate its Guarantee of Service (GoS) 2.0, an embeddable software suite that is used to manage network congestion in a controlled and predictable way. GoS incorporates patented queueing, rate-limiting and traffic-shaping mechanisms.

Tolly Group engineers examined the Quality of Service capabilities offered by GoS 2.0 versus Low Latency Queueing (LLQ), a QoS technique developed by Cisco Systems, Inc. to specify low latency behavior for a specific traffic class, such as real-time voice. GoS 2.0 allows network managers to define nine distinct traffic classes (plus Best Effort) using a 3x3 matrix assigning different degrees of latency and loss. Cisco provides for a special low latency queue, plus the definition of traffic assigned to several “Fair Queues” (plus Best Effort).

Tolly Group engineers subjected both technologies to a trio of increasingly complex QoS tests: a basic scenario, a differentiation scenario in which engineers measured the delay associated with streams of different quality in a congested network; and finally a simulated VoIP scenario where engineers analyzed the delay and amount of packets received from 25 VoIP calls and one Best Effort congesting stream.

In each scenario, GoS delivered predictable, uniform low latency, as opposed to Cisco’s LLQ, which offered consistently higher latency than GoS and less control in multistream tests with congestion. Tests were conducted during August 2004.