Decentralized Storage with IPFS: How does it work under the hood?
summary
Explains how IPFS turns files into chunked, hash‑addressed CIDs using IPLD and serves them via a Kademlia-style DHT (provider + peer records); Kubo is the common implementation and the daemon exposes an HTTP gateway. Measurement uses a custom crawler (nebula), global probe nodes and gateway logs, running ~9.5k crawls every 30 minutes; finds ~50% of nodes churn out within one hour and ~70% within four hours, so provider records use 20 replicas; ~97% of peers map to non-cloud hosts.
AI-written summary · about
chapters (18)
- 0:00Intro
- 0:28Dennis Trautwein
- 1:07Agenda
- 3:07Value Proposition
- 4:41Installation
- 6:59Content Identifier
- 8:33Chunking
- 11:57Daemon
- 13:14Challenge III
- 14:39Distributed Hash Table
- 17:01Bootstrapping
- 18:52Retrieving Content
- 20:12Publishing Content
- 21:07Content Lifecycle
- 22:43Measurements
- 24:29Peer Churn
- 25:36Cloud Provider Dependency
- 26:40Get Involved!
