Abstract

Cray Communications commissioned The Tolly Group to perform a comparison of key features of the MatchBox Router vs. the Cisco 2500 branch office router. The test included comparison of router performance (both with and without compression), interoperability tests, and an ease-of-use evaluation.

Results show that the MatchBox Router delivers equivalent performance to the Cisco 2500 in tests of raw throughput across high-speed, leased lines. In data compression tests, the MatchBox Router outperformed the Cisco 2500 by up to 18.5%. Additionally, the MatchBox Router demonstrated ease-of-use advantages over the Cisco 2500. The Tolly Group also verified that the MatchBox Router is interoperable with the Cisco 2500 in a number of common scenarios.

Summary:

This November 1995 Tolly Group report compares Cray Communications’ MatchBox Router with the Cisco 2500 for throughput, compression, interoperability, and ease of use. On raw routing over a T1 link, performance was essentially equivalent. In bi-directional tests with 64-byte frames, MatchBox reached ~2.57 Mbit/s for IP and ~2.58 Mbit/s for IPX versus Cisco’s ~2.61 Mbit/s (IP) and ~2.39 Mbit/s (IPX). With 1,500-byte frames, both saturated the full-duplex link at ~3.07 Mbit/s for IP and IPX. In uni-directional tests, both products saturated the link across frame sizes; 64-byte results appear to exceed T1 bandwidth because throughput was recorded on the LAN side and WAN header replacement reduces transmitted size. 

Compression & interoperability:

WAN compression favored MatchBox by up to 18.5%. On a 128 Kbit/s LAPB link, MatchBox achieved 2.43:1 for both IP and IPX versus Cisco’s 2.08:1 (IP) and 2.05:1 (IPX). At 64 Kbit/s results were similar: MatchBox 2.43:1 for IP, IPX, and bridging; Cisco 2.18:1 (IP) and 2.04:1 (IPX). Cisco lacked LAPB bridging, so bridging compression was not tested on the 2500. Interoperability over PPP was verified across IP Class C networks and subnets and IPX using Ethernet II, 802.3, and SNAP framing, including connect, login, and file transfers in both directions. 

Ease of use & methods:

MatchBox emphasized simpler deployment: auto-sensing WAN link type and speed, auto-enabling the LAN port, optional auto-loading configuration from a boot server, a menu-based configuration utility, telnet access, and a factory default IP for first login; Cisco supported telnet and auto-sensing speed but not the other listed conveniences. Tests used 10 Mbit/s Ethernet frame generators, measured maximum throughput at a 2% discrepancy threshold, and computed compression as effective throughput vs link bandwidth; routing used PPP and bridging used LAPB where supported.