Since Bluecurve released Dynameasure 1.0, the Windows NT Magazine Lab has been using the software to test system performance. Dynameasure has proved to be a very flexible tool for evaluating enterprise-level technology. Because each physical workstation can generate the load of 25 clients, the Lab needs only 40 systems to simulate a real-world workload of 1000 network users. This capability gives the Lab a realistic enterprise-level test environment that lets us quickly pinpoint an operating network's strengths and weaknesses. With Dynameasure Enterprise 1.5, the Lab can test both SQL Server and file-sharing performance.
The Lab reviewed Dynameasure Enterprise 1.5 in a network environment that includes a Digital Equipment Prioris HX 5133DP with 64MB of RAM, dual 133MHz Pentiums, and a 1GB hard disk as the Primary Domain Controller (PDC). We installed Dynameasure's control client and control server software on one machine: a quad 200MHz Pentium Pro IBM PC Server 704 with 256MB of RAM, a 500MB hard disk, a 2GB hard disk, and an Adaptec (four-port) PCI adapter. The servers ran Windows NT Server 4.0 with Service Pack 2. For connectivity, we conFigured a CISCO Systems Catalyst 5000 Series into four switches.
We connected 30 clients running NT Workstation 4.0 to four repeaters, which in turn connected to the CISCO. All the clients had at least a 100MHz Pentium and 32MB of RAM: Six Compaq Deskpro 510 workstations connected to a Compaq Netelligent 11108 100Base-TX Class II repeater, 7 Digital Venturis FX 5133 workstations connected to a Cogent S-1200 TX 100Base-TX Class II repeater, 10 PC-clone workstations connected to a second Cogent S-1200, and 2 HP Vectra LV Series 4 workstations and 5 DTK Computer workstations connected to a third Cogent S-1200.