RSS |
 |
| F E A T U R E | |
| |
Windows 2000: Worth the Pain (Almost) March 6, 2000 How We Tested Our test bed for file and application services included dual 500-MHz Pentium III servers with 512 MB of RAM and five UW-SCSI disks--one for the C drive, one for D and three for E in a striped set. One of the servers was running Windows NT 4 with Service Pack 5 and also served as a primary domain controller. The second server was running Windows 2000 and served as a domain controller. Our client machines were Pentium III 600s with 512 MB of RAM. Each client was connected to an Alteon WebSystems' Aceswitch 180 10/100/1000 switch at 100 Mbps, full-duplex. The servers were connected to the same switch via an Alteon ACEnic Gigabit Ethernet adapter.
Setting Up the Benchmarks The test scaled from 20 to 200 virtual clients in steps of 20, and ran on seven 600-MHz Pentium III Windows 2000 Professional machines (see File Server Comparison in PDF format). Each step included 5 minutes of ramp-up time, 3 minutes 45 seconds of execution time, 1 minute 15 seconds of ramp-down time and 30 seconds of quiet time. To perform the application server benchmark, we used Microsoft's Exchange Server 5.5 Enterprise Edition with Service Pack 3 and Microsoft's LoadSim program. Using the Microsoft Exchange Optimizer, we optimized Exchange for the server and prepared it for heavy loads by applying the specifications listed in Microsoft's Knowledge Base article Q234702, "MTA Queue to Information Store Processing Slowly." Using LoadSim, we simulated for a period of three hours the activity of 2,800 simultaneous heavy Microsoft Outlook users. Four-hundred users were simulated on each of seven 600-MHz Pentium III-based Windows 2000 Professional machines; this ratio was well within the parameters specified by the LoadSim documentation. Prior to running the test, we pre-initialized each of the client mailboxes with a populated folder hierarchy and calendar. During the test, we tasked the clients with a series of typical Outlook user activities: creating, sending, forwarding, moving, deleting and replying to mail; changing and deleting appointments and responding to appointment requests; browsing, creating and deleting folders; and journaling (see Exchange Server Comparison in PDF format). During the first 30 minutes of the test, the server was brought to a steady state. We used the next two hours of the test for computing the results, leaving the last 30 minutes for ramp-down.
Web Performance Prior to running the Web benchmark, we followed the tuning guide found at msdn.microsoft.com/workshop/server/feature/tune.asp for the Windows NT 4 server, and we followed a performance-tuning guide supplied by Microsoft for the Windows 2000 server.
The Web benchmark consisted of numerous tests, ranging from static Web pages to pages that contained server-side includes. For the tests with multiple network cards, the clients were evenly distributed between the server's network cards--that is, in the two-NIC test, half the clients used the IP address assigned to one network card, and the other half used the IP address assigned to the second card.
| |
|
PAGE: 1 I 2 I 3 I 4 I 5 I 6 I 7 I 8 I 9 I 10 I 11 I NEXT PAGE |
|
To conduct the file server benchmark, we used Client/Server Solutions' Benchmark Factory 2.0 software. We ran a mixed I/O test that included file and directory creation and deletion, sequential and random reads, and sequential and random writes. The writes were not cached, so the scores were much lower than they would have been in a cached environment.Best of the Web
Data deduplication: Declawing the clones
Data deduplication is emerging as a critically important new arrow in the storage administrator's quiver to answer hard questions about the increasing problem in storage growth costs.
Compression, Encryption, Deduplication, and Replication: Strange Bedfellows
One of the great ironies of storage technology is the inverse relationship between efficiency and security: Adding performance or reducing storage requirements almost always results in reducing the confidentiality, integrity, or availability of a system.
WAN Optimization Whitelists and Blacklists
Optimization is a fantastic way of saving money and creating really happy customers at the same time, but it doesn't work flawlessly for all applications.
WAN Optimization as a Managed Service: It's Not About the Cost
This insight examines how organizations outsourcing their WAN optimization initiatives to a third-party go about achieving their goals for application performance, reducing operational costs, and streamlining enterprise infrastructure.
