Why dosen't IPFS store the content itself in Kademelia DHT?

, ,
1

IPFS only stores the ā€˜peerIDā€™ of the content provider in KAD-DHT. Why not store the content itself in the DHT ? Are there any limitations with that approach ?

2

  1. Way to much data to store and keep synced in DHT. The required maintenance traffic would be immense.

  2. Reduced resistance to churn resulting in data loss.
    If the value of a PeerID are lost due to all peers responsible for that particular key going offline no big deal, that info can be rebroadcasted at will (and is rebroadcasted every 24 hours by default). If a block for ā€œsuperImportantData.zipā€ is lost to churn it is unrecoverable unless someone happens to have a copy not in the DHT, detects that it is missing, and rebroadcasts it.