The difference between these is who is initiating the conversation. In Polling, the monitoring server is initiating the conversation asking the network devices for information. Traps are the opposite where the network devices are sending information to the monitoring server right away when something happens.
The network device has an issue, let’s say an interface down, that network device sends that information to the monitoring server right away. Polling is set to certain amount of time. Depending on how the polling is set, you might not get the information right away. So if the server is set to poll the network device every 120 seconds which I believe is Solarwinds Orion default, there’s a possibility you won’t see the alert for 2 minutes.
An SNMP poll is initiated by the monitoring server like Solarwinds Orion and the network device (router/switch/firewall) responds to that server.
An SNMP trap is initiated by the network device (router/switch/firewall) when it has information to send (usually some event happened) and does not want to wait for the server to ask for information.