I started working with Firebase about a year ago after going to a Facebook Hackathon in Austin. I instantly fell in love with the system and don't have enough good things to say about it. The benefits with Firebase are obvious, but it can be daunting to use such a new system for sensitive information.
1). if Firebase disconnects a websocket, it stores the information and once the socket is reconnected, it wills end of the information. It is wonderful and has saved us quite a bit of headache.
2). It handles large amounts of data really well. I'm not 100% sure I follow your question, but it sounds like you want to ensure it can transfer/store the large amount of data. While we never had to send a ton of data all at once, we have pulled in large amounts (half a gig of text) and had no real issues (a bit slow for reporting, but that can't really be avoided).
3). If you want to store all the information locally as well, you can pass the info to Firebase via any backend that can ping a server (ie: curl).
Also, one of the awesome things that Firebase introduced a couple months ago was validation (so no one can ping Firebase and tell them to drop all your info without being validated). This uses read/write privileges that are really easy to set up.
Let me know if you have any more questions about it