HP Auto Port Aggregation Performance and Scalability White Paper Part number: 5555–1245 published February 2006 Edition: 1.
© Copyright 2006 Hewlett-Packard Development Company, L.P. Confidential computer software. Valid license from HP required for possession, use or copying. Consistent with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government under vendor's standard commercial license. The information contained herein is subject to change without notice.
Table of Contents ...................................................................................................................7 Introduction............................................................................................................................................7 Executive Summary.................................................................................................................................7 Benchmark Results..........................................................
List of Figures 1HP APA Throughput for Unidirectional Transmit and Receive Workloads............................................................8 2HP APA Throughput for Bidirectional Transmit and Receive Workloads..............................................................
Introduction HP Auto Port Aggregation is a software product that creates link aggregates, often called “trunks,” which provide a logical grouping of two or more physical ports into a single “fat pipe.” Two primary features are automatic link failure detection and recovery as well as support for load balancing of network traffic across all of the links in the aggregation.
Benchmark Results The HP APA link aggregates provided outstanding scalable performance for various workloads when used in accordance with the recommendations in “Benchmark Components” (page 11) and “Configuration ” (page 9). Link rate throughput is maintained for link aggregates with 2, 4, and 8 Gigabit Ethernet ports.
Bidirectional Throughput Performance Figure 2 shows HP APA throughput numbers with bidirectional transmit and receive workloads for 2-, 4-, and 8-port link aggregates. Figure 2 HP APA Throughput for Bidirectional Transmit and Receive Workloads 16000 14000 Megabits per Second 12000 10000 8000 6000 4000 2000 0 Bidirectional throughput 2-port APA 3682 4-port APA 7320 8-port APA 13524 Figure 2 shows that HP APA provides an exceptional sustained data rate of 3.68 Gb/s for a 2-port aggregate and 7.
throughput or the data rate for the networking work load, especially with link aggregates using four or more Gigabit Ethernet ports. • Set the load balancing algorithm for the HP APA aggregate on the server to ensure that the outbound traffic server is evenly balanced. Use the lanadmin command to set the outbound load balancing algorithm used by HP APA. HP APA provides a variety of load balancing algorithms to suit the individual needs of your configuration.
• Cumulative STREAMS patch PHNE_33313 or a superseding patch. This enables the STREAMS enhancements already in the TOUR 3.0 bundle for 11i v1. You can download this patch from the HP Information Technology Resource Center (HP ITRC) Web site at: http://www2.itrc.hp.com/service/patch/mainPage.do • If you have the HP IP Filter product installed on your system, upgrade to the latest version. You can download the latest version for free at: http://software.hp.
Clients (Systems That Drove the Test Load) Eight HP Integrity rx2600 clients, each with two 1.5 GHz Intel® Itanium® 2 processors with 6 MB L3 unified cache and 2GB memory. The clients ran HP-UX 11i v2. Each client used a single Gigabit Ethernet adapter to drive the test workload on the server. Gigabit Ethernet Adapters The rx8620 server had four HP A7012A, PCI-X 2-port 1000Base-T Gigabit Ethernet adapters placed in highest performing PCI-X 133 MHz slots.
Benchmarking Software HP generated test results using the netperf Version 2.4.1 benchmarking software and the following options: -C local_CPU_rate -c remote_CPU_rate -H remoteIP -I time -S socket_buffer_size -s socket_buffer_size -T localCPU, remoteCPU -t test_type Specifies to gather local CPU utilization for the test. You can calibrate the CPU rates separately using LOC_CPU and REM_CPU tests. HP used -C 1.5e+09 for our tests because both server and clients used 1.5 GHz CPUs.
