KVM Switches - 30 Mar 2000

Is a UTP-based solution worth the cost?

When you can't afford the additional space to have a monitor, keyboard, and mouse for each system you manage, a small keyboard/video/mouse switch (KVM switch) is the answer. Simple dual-computer, rotary-switch-activated KVM units sell for approximately $50. However, traditional KVM switches use three cables to connect each computer to the switch. Larger environments, such as the Windows 2000 Magazine Lab, might need many KVM switches, which means a huge number of cables. Space considerations aside, each cable costs $25 to $40. Wiring up a lot of computers can become a complex, crowded, and expensive procedure, even for a simple 12- or 24-system switch. So, the Lab also uses multisystem UTP connections. UTP-based solutions can cost thousands of dollars but are an incremental cost in a large KVM installation.

Our network testbed machines can have a KVM switch on the rack and need only local monitoring, so the less-pricey KVM solutions work well. But even the simple multicable solutions for a 60-machine test network can run close to $8000. The additional cost of a UTP-connected KVM switch is worthwhile because such a switch adds flexibility and ease of use. UTP solutions are much easier to implement than traditional multiple-cable connections: UTP connections use one Category 5 cable, which is cleaner to install than are the cables necessary in a more traditional KVM environment. To run UTP cable between our offices and the Lab, we're looking at two solutions. Raritan Computer's Cat5 Reach, an add-on to Raritan's KVM products, uses Category 5 cabling to support connections. Minicom Advanced Systems' Supervisor Phantom uses UTP to connect the target system's PCI card to the management console system. Initially, the Minicom solution required you to open the system to add the PCI card. But Minicom recently released its Phantom in a box solution, which moves the Phantom electronics to an external box that you attach to a target system. The solutions are comparable, although Raritan supports cable lengths up to 650 feet, whereas Minicom's connection lengths are limited to 360 feet.

These UTP-based solutions seem to avoid another problem (aside from the cable length limitation) that plagues cable-based solutions: video signal degradation on monitors running at higher screen resolutions. Because the Lab uses KVM switches to decrease the number of monitors we need in our test environment, we sometimes set screen resolutions up to 1600 x 1200 with 24-bit color. With resolutions starting as low as 800 x 600, we see a gradual degradation of the monitor's video image resolution on systems using cabled KVM switches. This degradation worsens as we raise the monitor's video resolution. Some of the higher-priced (approaching $100 per port) cable solutions provide decent quality video at resolutions of 1024 x 768 or even 1280 x 1024, but the limitations become apparent beyond that range. With the great amount of information available from modern systems management products, network management tools, and the control consoles that we use to run our Benchmark Factory benchmarks, we're constantly in need of more screen real estate. Therefore, the artificial limitations imposed by KVM switches are extremely annoying. With UTP-based solutions, the video at all our supported screen resolutions is as crisp as it would be on a locally attached monitor.

Easy connectivity and distance support make UTP an appealing solution to KVM's overcrowding problem. However, UTP products are expensive (adding approximately 25 percent to your overall installation cost), so the scale of your installation will have a significant effect on the cost.