Abstract

One-page summary of full report. Published in advance of the full report. 

3Com Corp. commissioned The Tolly Group to benchmark its SuperStack 3 Switch 4300 versus the Cisco Systems, Inc. Catalyst 3548 XL switch in competitive Layer 2 Fast Ethernet/Gigabit Ethernet forwarding performance tests. The Tolly Group evaluated the steady-state, zero-loss (less than or equal to 0.001%) bidirectional (full-duplex) frames per second (fps) throughput of both switches in a full-mesh configuration. In addition to the standard frame sizes, engineers benchmarked the throughput of both switches when handling “odd-sized” frames of 65, 129 and 257 bytes, in order to provide results for the full-spectrum of frame sizes. The 3Com SuperStack 3 Switch 4300 was tested with 48 Fast Ethernet ports (100Base-TX) and four Gigabit Ethernet ports (1000Base-SX). The Cisco Catalyst 3548 XL was tested with 48 Fast Ethernet ports (100Base-TX) and two Gigabit Ethernet ports (1000Base-SX). Both devices were configured with 10 Fast Ethernet ports meshed with each Gigabit Ethernet port, and the remainder of the Fast Ethernet ports meshed with each other.

Summary:

The Tolly Group conducted a competitive performance evaluation in early 2001 of three Layer 2 Gigabit Ethernet switches: the 3Com SuperStack 3 Switch 4900, the Cisco Catalyst 3508G XL, and the Cisco Catalyst 4003. The focus of the testing was to measure zero-loss, bidirectional throughput at various Ethernet frame sizes (64, 512, and 1,518 bytes) and compare performance against each vendor’s stated specifications. The evaluation was sponsored by 3Com and performed using a Spirent SmartBits traffic generator in a full-duplex, full-mesh environment.

The results showed the 3Com SuperStack 3 Switch 4900 delivering 100% wire-speed throughput (32 Gbit/s) at all frame sizes tested, matching both its theoretical and advertised performance. In comparison, the Cisco Catalyst 3508G XL achieved only 50% of its theoretical throughput and 80% of its advertised capacity, while the Catalyst 4003 performed even lower—delivering only 15–20% of its theoretical maximum and around 42–56% of its advertised throughput, depending on frame size. Notably, the 3Com switch achieved full wire-speed performance on all 16 ports without packet loss, while Cisco’s switches exhibited clear throughput limitations.

The analysis highlights that the 3Com switch offers better price/performance value, especially for networks utilizing copper-based cabling to achieve Gigabit Ethernet. Its non-blocking architecture, support for both 1000Base-T and 1000Base-SX ports, and strong Layer 2 feature set make it particularly suitable for high-performance data centers and backbone environments. The Tolly Group concludes that the 3Com SuperStack 3 Switch 4900 stands out as the only device in the comparison capable of full wire-speed switching across all ports under maximum load conditions.

Note: 3Com was acquired by HP (later HPE) in 2010.