The Replayer depends on MongoDB (i.e. the Standalone Repository's database) and RabbitMQ (for queuing during dispatch). The correct versions and configurations of both will be supplied by installing the Data Hub and the Standalone Repository before installing the Replayer. The Replayer is designed to be run side-by-side with these other PreEmptive Analytics products on the same machine.
If you wish to use the Replayer on a different host than MongoDB and/or RabbitMQ, the connection settings for both are configurable in the
Replayer.exe.config file, in the installed directory.
The Standalone Repository and Replayer together require the following minimum system resources:
Disk space requirements are subjective. Consider the worst case: if you replay all the data stored in the Repository's database, and the destination is either offline or returning an error response for every envelope, the Replayer would store the envelopes in an offline or error queue respectively. This can cause Replayer to use as much as double the disk space as is already in use by the Repository. (The Repository compresses data for storage, but the Replayer does not.) However, as long as the data is being delivered successfully, the Replayer will not queue envelopes to disk.
In most cases, the output throughput of the Replayer will be limited by either the CPU of the Replayer host, or by the destination's ability to accept envelopes (which is usually itself CPU-limited by that host). A host meeting the above recommendation can typically sustain 400 envelopes/second output.