Understanding EMG throttling Print
Thursday, 14 July 2011

When EMG limits throughput, either due to license restrictions or due to a max throughput configuration, it throttles messages imposing a "soft" limit. If messages are sent to EMG more rapidly than the license allows EMG will introduce a slight delay before sending back responses to the message submit operations.

Since the client application can only send a limited number of messages (usually 10 or so) before waiting for response this will in effect slow down the throughput.

No messages will be dropped in EMG but rather a decision has to be made on the client side what to do if messages are generated more rapidly than they can be sent out. Normally this is not a problem if this condition is temporary but if longer term then it can cause a large queue on the client side and unacceptable delays in message delivery.

As an example if a specific connector limits throughput for incoming messages via THROUGHPUT=5 (max 5 messages per seconds), EMG will delay the response to the client on average 0.2 seconds for each message send operation.

EMG only limits incoming mps by license. Since EMG itself does not generate any messages this will in fact mean that messages will not be sent out more rapidly than they are received (incoming mps >= outgoing mps for EMG).

Throttling is not applied to delivery reports.