Abstract

Tolly Group report #202158 (December 2002) documents a converged-network benchmark in which Extreme Networks commissioned Tolly to pit its BlackDiamond 6808 core plus Summit 48si access switches against two configurations of Cisco Catalyst 6509 (Gigabit line cards WS-X6516 and WS-X6816). Engineers built a micro-cosm IP campus, enabled QoS, and then ramped background load on core and inter-switch links from 30% to 100% to see how well each fabric protected VoIP and mission-critical data.

With QoS active, Extreme kept toll-quality voice (PSQM 0.4), sub-100 µs latency and ≤ 3 µs jitter at every load point—even at 100% link saturation, latency was only ≈ 80 µs. It simultaneously forwarded 100% of mission-critical frames, whereas Cisco began to falter as soon as congestion topped 40%. At 40% load the Catalyst 6509/WS-X6516 configuration’s voice score collapsed to PSQM 5.6 with 379 µs latency and 21.8 µs jitter; by 50% load it bottomed out at PSQM 6.5 and 1.4 ms latency, forwarding only 32.6% of high-priority packets. At the severest 100% load, Extreme still hit PSQM 0.4 and moved every critical frame, while Cisco’s two variants dropped between 31% and 77% of those packets and stretched latency past 0.9–1.6 ms.

Cost analysis amplified the technical gap: the Extreme solution (64 GbE core ports + 2 access switches) priced at $137,605, whereas the comparable Cisco designs cost $206,904 (WS-X6516) and $226,904 (WS-X6816)—roughly 50–65% higher. Combined, the study shows that under heavy congestion only Extreme’s BlackDiamond/Summit architecture sustained consistent toll-quality VoIP, low latency/jitter, and perfect delivery of mission-critical traffic, while doing so at a significantly lower price-per-performance point than Cisco’s AVVID infrastructure.