A policy engine hub can have multiple event queues, configurable by email size, with each queue serving multiple policy engines. Distributing emails across multiple queues in this way allows fast‑track processing for emails of different sizes. Registry settings on the hub host server define the queue size bands and specify which policy engines are assigned to each queue—see Hub event queues for details. When a policy engine has finished processing an email, it is passed back to the hub before being returned to the source application.
Policy engine hub architecture
A policy engine hub (1) can accept emails from various sources, including Event Import (2a) and an email server agent (2b).
Emails arriving at the hub are added to the input queue (3). The hub then assigns each email to an event queue (4). A hub can have multiple event queues (three in this example), configurable by e‑mail size. Registry settings on the host machine (1) define the size band for each queue.
Each queue can be served by multiple policy engines (5). Registry settings on the host machine (1) specify the policy engines allocated to each queue. To minimize processing times, if a queue is empty then idle policy engines assigned to that queue are also permitted to poach messages from other queues (6), but only from queues with a smaller maximum size limit.
After a policy engine has successfully processed an email, the e‑mail is passed back to the completion queue (7) on the hub before being finally returned to the source application (2a or 2b).
|
Copyright © 2014 CA.
All rights reserved.
|
|