Understanding architectural change in Tie-Breaker machine requirement upgrading from AutoSys 4.5.x to R11 release.
For AutoSys 4.5.x a "Tie-Breaker" or "Third machine" is necessary for DUAL EP mode even if the instance has only one database (Event Server). In 4.5.x the Primary and Shadow Event Processors monitor for a DIBS file on the third machine. The Primary EP is constantly pinging the Shadow EP and checking for the DIBS file. If that process is interrupted or if the Shadow EP notices it has not received a Shadow ping from the Primary EP it will connect to the Third machine and attempt to grab/create the DIBS file. Which ever EP gets to the 3rd machine 1st and can create the dibs file will remain up and assume control for processing events. The other EP will stop.
For AutoSys R11.x this has changed. You only need a "Tie-Breaker" machine if you are in HADS (High Availability Dual Server) mode. High Availability mode using both a Primary and Shadow Scheduler and database. AutoSys R11 no longer uses a DIBS file for failover control. Instead the schedulers use tables in the database to make each other aware of their state, role and status. They constantly read and update the tables accordingly. The Tie-breaker Scheduler also connects to the databases and resolves contentions and eliminates situations in which one Scheduler takes over because its own network is down.
For more details please review the Unicenter AutoSys Job Management Implementation Guide for UNIX/Windows R11.x