ET/BWMGR Failover System

Emerging Technologies bandwidth manager appliances can optionally be equipped with a hardware "failover" module that can eliminate the "single point of failure" concerns with this type of product, as well as enhance the maintainability of the product. The diagram below illustrates the setup of the appliance when configured with the failover option. One Ethernet port is connected to your network normally. If you will be managing the system from internal devices, then the management port should be connected to the local network switch (or hub). If you will be managing it from a remote location (ie via the internet), then you can connect the management port to the other "side".

When powered down, the appliance is in the "failover" state, so traffic will flow when wired properly. Even as the machine is powered up, the ports remained connected until the ET/BWMGR software is ready to process traffic. Then the ports are "enabled", at which point they act as 2 independent ethernet ports. Traffic is then bridged by the bandwidth management software normally, in the same way it does if you are using 2 regular ethernet cards without hardware failover.

The appliance monitors its "health" via a software watchdog timer, which must inform the failover hardware periodically that it is in good working order. If the failover mechanism does not get this "message" (which could be caused by the machine locking up or a software or OS failure), it will then connect the ports allowing traffic to flow on your network as if the bandwidth manager appliance wasn't there.

You might ask why the "management" port is necessary, and the answer is that it isn't, but it adds some functionality that can be of great use. For example, you can put the appliance into failover to do system maintenance or when doing system upgrades, allowing you to safely perform such tasks remotely without concern that your network will be down if something goes wrong. It also allows you to manually take the machine out of failover in the unlikely case that a "false failover" occurs.