How we tested Cisco's Catalyst switch

Network World - We assessed the Cisco Catalyst 4500 with tests of functionality, performance and Ethernet energy efficiency. The performance tests used the Spirent TestCenter test instrument to measure layer-2 unicast throughput and latency; layer-3 throughput and latency, with separate IPv4 and IPv6 tests; layer-2 multicast throughput and latency; layer-2 MAC address capacity; and in-service software upgrades and downgrades (ISSU/ISSD). All tests used version 03.02.00.2.70 of Cisco’s IOS-XE software (with a .80 release for the ISSU test) and version 3.62 of Spirent TestCenter software.

During all performance tests, the Sifos PowerSync analyzer drew maximum power of 60 watts at each of 24 switch ports in a single module. There was no impact on switch throughput or latency from the power draw, as evidenced by virtually identical performance for streams on ports with and without power over Ethernet.

We assessed UPoE functionality in two ways: First, by using the Sifos analyzer to draw 60 watts concurrently on 24 switch ports, and second, by attaching various devices to the switch. These devices included a Cisco 9971 phone with embedded thin-client computer; a Samsung SyncMaster NC220 23-inch monitor; and a BT ITS.Netrix "turret phone."

The unicast performance tests involved a fully meshed traffic pattern among 384 gigabit Ethernet switch ports. The Spirent TestCenter generator/analyzer offered traffic for a duration of 60 seconds, and measured the latency of every frame received. We repeated this test with 64-, 256-, 1,518-, and 9,216-byte Ethernet frames (and substituted 78-byte frames for 64-byte frames in the IPv6 test case, since this is the shortest length possible that can accommodate the signature field used by the test instrument).

The multicast traffic test involved a traffic pattern with one transmitter port and 383 receiver (subscriber) ports. Here, all 383 receiver ports on the Spirent TestCenter instrument joined the same 1,000 multicast groups using IGMPv3. After the switch's IGMP snooping table was fully populated, the test instrument then offered traffic to the single transmit port, with destination addresses of all 1,000 multicast groups. As in the unicast tests, the instrument measured throughput and latency for four frame lengths.

To determine media access control address capacity, engineers used the RFC 2889 MAC address caching test on three ports to determine the largest number of MAC addresses the switch would learn without flooding.

To measure ISSU/ISSD, engineers configured Spirent TestCenter to offer 256-byte frames to the switch at line rate in a fully meshed pattern. While traffic was running, engineers initiated an upgrade of the system software to a new version, and derived conversion time from frame loss at the end of the test. Engineers then repeated the test by downgrading the switch to the original image, and again derived cutover time from frame loss.

We used a clamp meter to measure the switch's energy-efficient Ethernet (EEE) capabilities. This test compared energy draw when idle with and without EEE enabled.

Return to main test.

Read more about lan and wan in Network World's LAN & WAN section.