Currently my dumb ISP would block my Internet if there are too many TCP/UDP requests, from legit P2P services (e.g. P2P software updates), to video-conferencing apps (Webex, Zoom, Skype).
If I use a VPN, under the hood, will all the connections bundle under one giant VPN connection, or will each TCP/UDP request spawn independent VPN connections? i.e. can the ISP meta-analyze my requests over VPN (# requests, packet size etc.) to block my Internet?
OpenVPN, ZeroTier and WireGuard all use UDP, but on a single port.
All your VPN-borne traffic will be encrypted and encapsulated using the VPN’s protocol and port. No matter what the protocols, destination IPs and ports you’re accessing through the VPN, your ISP will still only see traffic on a single port and protocol between you and your current VPN server.
Your ISP won’t (generally) be able to determine what you’re using the VPN for, but traffic analysis of the encrypted VPN packets’ sizes and timing _is_ surprisingly accurate, especially when the VPN is only being used for one purpose.
ping is generally easy to spot, as are DNS requests, along with simple HTTP(S) sessions etc. Streaming video looks different (lots of incoming, full-size packets), and – as you suspect – P2P has fairly distinctive packet characteristics too.
Can they _absolutely_ prove you’re using P2P over a VPN? No.
If you deliberately have multiple activities using the VPN at the same time, it makes it much harder to detect a specific type of activity. Watch a long video over YouTube whilst P2P’ing, for example.