We tested these remote-control products at our Syracuse University Real-World Labs®. Performance tests determined screen-refresh rates between masters and hosts by transferring bitmap images during remote control sessions. File transfer performance was tested by transferring document files between masters and hosts. To grade each product, we combined the raw performance with an analysis of the features. We also tested for error recovery and detection and verified that clear text was not passed in the authentication and the data streams during encrypted sessions.
To test features and functionality, we installed, where necessary, NOS support. For performance testing, we used two Dell Computer GX-1 workstations (500 MHz). The GX-1s ran Microsoft Windows 2000 Professional. When possible, we turned off caching. When that was not possible, as with Web browser remote control, we stopped and started sessions and cleared cache between sessions. We inserted a Shunra Storm STX-100 between the workstations to emulate traffic at 56 Kbps--the slow speed provided a level playing field for testing screen-refresh and file-transfer capabilities.
For screen-refresh testing, we used a pallettized bitmap file with 8-bit color depth (223 KB) and three true-color RBG bitmaps at 24-bit color depth to transfer between master and host computers. For file-transfer testing, we used document files that ranged in size from 297 KB to 5,229 KB in several formats. Each file was transferred between guest and host three times with and without encryption.
For the most part, we found no noticeable difference in performance for encrypted and nonencrypted sessions. Because each program provided various options for exchanging keys and encrypting data, we disabled encryption to put each product at the same starting point. However, the products from Novell, Spartacom Technologies and uRoam defaulted to encrypted sessions, but this did not seem to hinder their performances.
To determine error recovery and detection, we unplugged the two test computers from the Shunra Storm and connected them to a 100-Mbps hub. We started a remote-control session and a file-transfer session for each program and unplugged the host from the hub. We did this three times for each product and averaged the three tests.
To test whether these applications encrypted the authentication and data stream to prevent clear text from passing on the wire, we used WildPackets EtherPeek NX 1.0.0 to capture and analyze all packets transferred between guests and hosts while connected to a 100-Mbps hub. For the Novell, Spartacom Technologies and uRoam products, we added the server into our packet-capture sequences.
Finally, we looked at the memory usage on hosts to determine the cost of remote control on corporate desktops and servers. We were careful to compare apples with apples in the memory requirements. For both Altiris and Novell ZENworks, we included the base client footprint and the remote control applications. Other components of the packages were not included in the calculation. Memory usage was measured while the programs were in memory and dormant (no remote-control session active) and active in a remote-control session.
RSS
