If you have some more time i would recommend watching this presentation of IPFS:
https://www.youtube.com/watch?v=h73bd9b5pPA (you can also checkout other videos posted on
https://ipfs.io/ )
If you don't have that much time this equation can tell you what ipfs is:
IPFS = HTTP + SFS + GIT + BitTorrent + DHT + ...
Switching from http/ftp mirrors to ipfs mirrors would give following benefits:
1) ability to download package from multiple mirrors same time
2) ability to download package from other peers downloading same file (like BitTorrent)
more users downloading = greater download speed for everyone
3) build-in versioning (like git)
4) ability to download small patch/diff data instead of downloading whole new version of file (like git pull)
5) build-in data deduplication
6) man-in-the-middle resistant (hash checking as part of protocol)
7) Many other small features not listed here
IPFS can be also mounted (via FUSE) and used like any other unix filesystem
When mounted as filesystem files and directories are lazy downloaded on first attempt to access given file/directory
In the future this can be potentially used to install linux applications/libraries on the fly when user tries to run them for a first time (and make rpm/deb deprecated)
You can already boot entire linux system from ipfs if you want.