You can't run multiple Event Servers of the same version on Windows. You can use the multi-process Event Server.
What you need depends on the number of watches, data size, data volume per time period, complexity of data, complexity of map and rules, functions used, 3rd party adapters used etc.
If you are limited to Intel hardware, and want the best environment to run DS TX, use RedHat AS 3. or HPUX on Itanium.
Get as much RAM as will fit in the box. Get as many actual CPUs as affordable.
Best bet is to design a load balancing / failover system.
Have one Event Server running on one bax to watch for triggers. This does not process the data. It will send the data via WebSphere MQ to a router map on one of the servers. It can do this round robin (using a rule and getandset) or you can have a script monitoring the server loads and updatings a DB table, so that the rule sends the data to the least busy server. Another script monitors a heartbeat map in each Event Server install on any server. If it is missed, it calls a recovery routine, if that doesn't work, that Event Server ID is taken out of the DB table for the round robin. If you servers are big enough, it could start a copy of that Event Server instance up on another machine. If the main distribution ES goes down, it would be started on another server. Servers can be added to the system with a small change to the script and DB table data.
With MQ clustered and your DB clustered, you have a fault tolerant, error tolerant, expandable system.
E-mail checks to me as required
BocaBurger
<===========================||////////////////|0
The pen is mightier than the sword, but the sword hurts more!
