What infrastructure should I use to persist messages from a public websocket?

I want to connect to a publicly available websocket, create a request and persist the subsequent, long running stream of events for later processing.

I will likely use node.js or Java to connect to the websocket and raise the request.

I want to avoid missing any messages that the service sends, so always on, re-connection and redundancy are important so that I can build an optimal data set at a later stage.

What AWS architecture would you suggest?