Abstract

Tolly Group Competitive Evaluation #202159 (December 2002) compared 3Com’s purpose-built SuperStack 3 Switch 4924, a fixed 24-port copper Gigabit Ethernet Layer-2 switch, with Cisco’s modular Catalyst 4006 (Supervisor III engine) and HP’s Procurve 4108GL in a full-mesh, 24-port, bidirectional SmartBits test bed. The goal was to confirm whether each platform could sustain wire-speed traffic across all packet sizes in high-density server-farm and desktop-Gigabit designs.

Performance results were unequivocal.

• SuperStack 3 maintained 100 % of theoretical throughput, 48 Gb/s aggregate full-duplex, at every tested frame size (64, 512 and 1,518 bytes).

• In the same tests, Catalyst 4006 delivered only about 25 % line rate (≈ 12 Gb/s) and Procurve 4108GL achieved roughly 38 to 42 % (18 to 20 Gb/s), leaving the 3Com switch with approximately four times Cisco’s throughput and 2.5 times HP’s.

• At the smallest frame size, SuperStack 3 forwarded 71.4 million frames per second versus 17.9 Mfps for Cisco and 27.1 Mfps for HP, underscoring its non-blocking design.

Operational take-aways. Because the 4924’s fixed configuration attains wire-speed without tuning, it offers plug-and-play deployment anywhere in the network while doubling, or even quadrupling, competitors’ throughput in the same 1 GbE copper footprint. For organizations planning dense Gigabit links to servers or desktops, the study demonstrates that the SuperStack 3 Switch 4924 can sustain full performance under worst-case loads, giving designers confidence that bandwidth assumptions will hold true in production.