How HTTP/2 Works

Three Main Problems With HTTP/1.1

Performance: Doesn't use full bandwidth
Complexity: Web applications have to optimize with multiple connections due to HTTP/1.1's lack of paralyzation.
Efficiency: Optimizations create device and server resource consumption

Main Benefits of HTTP/2

Speed: Improvements in actual and perceived latency.
Security: TLS is required by almost all browsers.

Features of HTTP/2

Request multiplexing: Uses a single connection to do multiple requests simultaneously.
The client can indicate priorities of requests and resources, and the servers can provide those faster.
The TCP connection is maintained for multiple seconds after the last request before it's closed, so additional requests are fast.
All data is binary. This means HTTP/2 is not backwards compatible.
Headers are optimized. Headers between requests are diffed so they're not repeated.
Servers can push objects without an explicit request. The client then caches those objects.
HTTP/2 keeps the semantics of HTTP/1.1.
The browses will take care of the optimizations for you with HTTP/2.

Major Problems With HTTP/2

TCP is slow.
If packet loss rises above 2%, all HTTP/2 benefits are lost.
HTTP/3 will solve these problems.

HTTP/2 on Backend Servers

Use the latest backend server and it'll likely be available. All major servers currently support HTTP/2.

ALPN: Application Level protocol Negotiation. This is a way for the client and server to agree on a protocol.

Side note: Protocol buffers are a GRPC construct that transmits data in a binary format that's an alternative to JSON. Most of the performance benefits come from this.