How we did it
 
|
|||
 | |||
|
|
|
|||
| |||
|
Since the x440 is meant to be a compute server (a server that handles computational tasks, cgi-scripts, database operations, and the like - typically not file and print services), we developed a test to maximize processor utilization. SSL session creation is a great way to utilize the processors because of the amount of math processing needed to pass the cryptography keys. To create the SSL transaction load, we used the WebAvalanche tester from Spirent Communications.
The WebAvalanche tester was configured to request a secure Web page loaded on the x440 running Microsoft Windows 2000 Advanced Server with Internet Information Services (IIS). The file requested is a 51-byte file to minimize the amount of network utilization during the test. To maximize the load on the server's processors, all session caching is disabled so the SSL keys are calculated with each Web request.
Advertisement: |
The four fast Ethernet ports of the WebAvalanche are connected to four ports of a Cisco 2948. The server is connected to another fast Ethernet port of the Cisco 2948. All Ethernet ports in the test network were configured for 100M bit/sec and full duplex operation.
To reveal the processor scalability of the x440, the number of attempted SSL transactions per second originated by WebAvalanche was increased until transactions were not completed because of saturation of the x440 processors. The processor utilization was monitored using the Windows 2000 processor utilization utility. The maximum number of successful SSL transactions without transaction failures was recorded. This methodology was repeated for two and four Xeon DP and two, four, six and eight Xeon MP processor configurations on the x440.
Since these benchmarking tests were designed to apply the greatest amount of stress possible to the server processors, the test results should in no way indicate the number of SSL transactions or users that the server can support in a real environment. It would be safe to say that the number of supportable "real" users would likely be much higher than our metrics.
Back to the main review: IBM eServer xSeries x440
RELATED LINKS
