How We Tested ISDN Failover Devices
We tested these three leased-line failover units using Cisco Systems 1601 routers with V.35 plug-in modules. Two test networks were created. The routers and external HTcomm CSU/DSUs were connected via a DDS (digital data service) crossover cable to simulate a leased line. One of the CSU/DSUs was designated as the master clock and the other a slave. Dell Dimension Pentium computers running Windows NT 4.0 were placed on each segment
, constantly engaged in TCP/IP pings and file transfers to and from workstations on the opposite segment.
The ISDN failover units were placed between each CSU/DSU and the router. ISDN BRI lines were connected as the failover circuits. A series of tests was then conducted to verify proper operation and timing of the failover and restore. These tests included disconnecting the DDS crossover link from each CSU/DSU to indicate leased network failure at different intervals, ranging from about one second to several minutes. The CSU/DSUs and routers were then powered off at different times to measure the failover units' ability to recognize loss of DTR (Data Terminal Ready), RTS (Ready to Send) and any other signaling each unit was capable of, such as "all ones" and "all zero." The units were also powered off during normal operation to test their ability to passively transmit signal from DCE to DTE in the event of power failure or unit processor failure. In addition, their ability to recover from power loss and ISDN circuit or remote unit failure and recovery were tested.
Each failover unit was tested as the initiating and receiving failover device and set at different failover intervals, ranging from immediate failover and line restoration from one second to 10 minutes. The Controlware IDB-64/2 and HTcomm LineSaver, because of their ability to block failover during certain time periods, were tested by initiating signal loss at intervals before, during and after our programmed blocking periods.
To test these failover units with bandwidth on demand, we initiated a failed circuit and flooded the circuit using Ganymede Software's Chariot benchmark. The line utilization parameter was programmed at 50 percent utilization for 30 seconds at up to 512 Kbps.
|
RSS
