Polices are pushed out to the agent for any of the following reasons

1. The Endpoint Agent connects to the Endpoint Server after being disconnected from the network.
2. The Endpoint Agent connects to the Endpoint Server after being restarted.
3. The Endpoint Server is restarted.
4. The SymantecDLPDetectionServer is restarted.
5. A policy is changed, enabled or disabled, then policies will be distributed.
6. A response rule changes, then response rules will be distributed. 

Starting in version 10, the policies will be compressed before being sent to the Endpoint Agent to reduce the bandwidth requirements.

If you believe that the network issues happen because of policy distribution, then you might try adjusting the following properties in Aggregator.properties on the Monitor:

data.flow.shipping.thread.pool.size = 5

data.flow.structured.thread.pool.size = 10

This step will cause policy changes be sent to fewer agents concurrently.  Thus, it will reduce the load on the network.  However, the time taken to distribute policy changes will also increase.