How we tested the various network compression devices.We created a midsize corporate network to conduct our testing. The main-office and branch-office networks were composed of\u00a0Cisco\u00a0Catalyst 2924\u00a0switches, Cisco 5500 Layer 3 switches and Cisco 7000 routers.frame relay\u00a0cloud configured in a hub-and-spoke topology, while the main office was connected to the Internet via a 10M bit\/sec Ethernet connection. This design let\u00a0LAN and\u00a0WAN\u00a0appliances be tested with minimal reconfiguration.Branch offices were connected to the main office through a T-1\u00a0Each network consisted of multiple Windows 2000, Windows XP and Linux servers and workstations. All three offices were equipped with Cisco 7960 IP phones.Baseline throughput and packet-loss testing were conducted using\u00a0Ixia Communications'\u00a01600\/1600T cards and\u00a0Spirent Communications'\u00a0SmartBits 2000 and 6000B cards. Each card was configured for full-duplex, 10M bit\/sec operation and used line-rate Ethernet frames between 64 and 1,518 bytes.Further baseline testing was conducted using\u00a0iperf\u00a0to push varying amounts of TCP and\u00a0UDP\u00a0traffic across each branch-office network. It also was used to test quality-of-service features by introducing congestion significant enough to prevent voice-over-IP calls from working over the branch-office T-1 lines. Subjective evaluations were made of voice quality before and after policies were put in place.FTP transfers of compressed and uncompressed files were initiated across each device to test its caching and compression capabilities. Each of the compressed and uncompressed files also was transferred via iperf for use in correlating test results.All tests were conducted in the\u00a0Network Services Interoperability Lab\u00a0at the University of Florida in Gainesville.Back to main review: "Supersizing existing WAN connections"