Better control over snmp polling for big network devices

ITNM has two different strategies for polling interfaces depending on the use of a polling filter:

- Strategy 1 (when no polling filter is defined): poll all OIDs for each interface.

- Strategy 2 (when a polling filter is defined): group multiple interfaces and do requests per OID.

For strategy 2, it means that if you have for instance 100 interfaces, and wanted to poll 2 OIDs, it will split the 100 interfaces into groups of 30 and then request OID1 for each group (with 30 varbinds on it) and then do the same for the second OID, etc. This however brings issues when a device has hundreds of interfaces and is not optimized for this kind of request (multiple interfaces per OID). The device may take a long time to answer each request and this causes timeouts in ITNM.

We would like to be able to finetune the polling in those situations, either by:

1) Deciding on which strategy to use even if a polling filter is defined

2) Deciding on how many parallel polling requests are triggered before next requests are made for same poll

3) Setting a custom polling timeout per device (so slower devices would not generate an alert while faster devices can continue to work as usual)

4) Have the option to set a delay between poll requests

These would allow to accomodate big devices and avoid timeout issues.