622 Adapters

ManualsBrandsHP ManualsNetwork HardwareHP-UX 11i v2 I/O Cards

Performance Report on ATM adapters 1

HP ATM 622 and 155 Mbps Adapters for HP 9000 Enterprise

Servers and Workstations

Performance Report

Product Numbers J2468A, J2469A, J2499A, J3573A, and J2806CA

Introduction

A series of tests were run to evaluate the performance of HP ATM 622 and 155 Mbps adapters

on various HP 9000 server and workstation platforms. The tests also measured the effect of

Classical IP and LAN Emulation interfaces, extended packet sizes, and TCP configuration

parameters on the adapter’s performance.

A second set of tests evaluated the scalability with respect to number of adapters and number of

CPUs on various HP 9000 server platforms.

This document presents the performance results and provides tuning guidelines to achieve

maximum possible performance from the ATM adapters.

Test Methodology

HP’s Netperf version 2.1 was used as the test tool in all the measurements. Netperf does the

measurements at the user level which is more conservative than measuring at the link level.

For tests with single adapter configuration, two target systems were used. A K580 system with

four CPUs and a K380 system with four CPUs were used as the target systems. For Multiple

adapter tests, up to four target systems were used. The target systems were configured with 1

Gbytes of memory. The target systems and the system under test were connected using the

FORE ASX-1000 ATM switch. The ATM software used was G.11.00.01 with Build Revision 6.21.

HP-UX 11.0 was used as the underlying operating system. All the configuration parameters were

default with the exception of modifying the LANE MTU size.

Single Adapter Configuration Test Results

The theoretical maximum TCP throughput is approximately 135 Mbps for the 155 adapter and

542 Mbps for the 622 adapter. The theoretical maximum throughput numbers are derived by

calculating protocol overheads at various layers.

With one ATM 622 adapter, the theoretical maximum TCP throughput is achieved in the inbound

direction. However, the maximum outbound throughput is limited to around 370 Mbps, due to how

the adapter uses the HSC I/O Bus. The bidirectional throughput is limited to 584 Mbps due to the

same reason. See Table 1 below.

Table 1: Maximum TCP Throughput on K580 (One CPU , Classical IP Interface, MTU 9180 bytes)

K580

(155 Mbps Adapter)

K580

(622 Mbps Adapter)

Outbound Throughput Mbps

134.55 370.1

CPU Utilization %

8.14 35.22

Inbound Throughput Mbps

133 540.4

CPU Utilization %

10.92 85.08

Bidirectional Throughput Mbps

261 584.5

CPU Utilization %

22.8 95.86

Summary of content (6 pages)

PAGE 1
HP ATM 622 and 155 Mbps Adapters for HP 9000 Enterprise Servers and Workstations Performance Report Product Numbers J2468A, J2469A, J2499A, J3573A, and J2806CA Introduction A series of tests were run to evaluate the performance of HP ATM 622 and 155 Mbps adapters on various HP 9000 server and workstation platforms. The tests also measured the effect of Classical IP and LAN Emulation interfaces, extended packet sizes, and TCP configuration parameters on the adapter’s performance.
PAGE 2
The packet size (that is, the Maximum Transmission Unit [MTU] size) influences the performance of the adapters. Table 2 below shows the maximum TCP throughput and CPU utilization figures for four different interface types for 155 Mbps adapters. Table 3 below shows the maximum throughput and CPU utilization figures for the same interface types for 622 Mbps adapters.
PAGE 3
The Request/Response performance of the adapter is measured in transactions per sec. A transaction is defined as the exchange of a single request and single response. A transaction consisting of request and response size of one byte each indicates the number of packets the adapter can process in a second. This transaction is done using the TCP protocol. Test results show that the single byte request/response performance is limited by the CPU.
PAGE 4
Multiple Adapter Configuration Test Results For systems with two HSC IO subsystems, the maximum throughput in outbound, inbound, or two-way direction is limited to 600 Mbps due to how the ATM adapter uses the HSC I/O Bus. For systems with four HSC IO subsystems (for example, the K580) the maximum throughput is limited to 820 Mbps in outbound, inbound, or two-way data flow.
PAGE 5
The MTU size makes a difference in the scalability. See Table 6 below. Table 6: Maximum TCP Throughput and Interface Type on K580 (four CPUs, six 155 Adapters) Outbound Mbps Inbound Mbps Bidirectional Mbps Classical IP MTU 9180 bytes 654 589 666 LANE MTU 9218 bytes 611 589 584 LANE MTU 1516 bytes 247 252 249 The maximum throughput and request/response performance on the K580 is shown in Table 7 below. The figures compare the performance when using six 155 adapters versus four 622 adapters.
PAGE 6
Tuning Guidelines • System-related Guidelines For the D-class servers, HP recommends the adapters be installed in the Turbo slot. For the K-class servers, HP recommends the adapters be installed in the expansion I/O board, avoiding the core I/O. When installing multiple adapters, make sure the adapters are evenly distributed on the HSC IO subsystems. This ensures efficient HSC bus utilization and better scalability.