The Principles Behind IPFS

Published in

MOAC

4 min readJul 18, 2018

IPFS’ ‘lofty’ goal is to possibly replace HTTP, so how does IPFS work?

IPFS assigns a unique hash value to every file (a file fingerprint created based on the contents of the file). Even if two files only differ in one bit, their hash values will still be different. So IPFS locates files based on their content, and not according to the traditional HTTP protocol’s domain names.

IPFS removes duplicate files from across the entire network and manages file versions, meaning that the change history of each file is recorded (this is similar to version management tools like Git and SVN). This makes it easy to check data saved on previous versions of a file.

The IPFS network queries files according to the file’s hash tag (unique across the entire network). Querying is very straightforward due to every file’s hash value being different.

It would be inconvenient to only transmit files according to their hash values, because hash values are difficult to remember, just like IP addresses. So mankind invented domain names. IPFS utilizes IPNS to reflect hash values as an easily remembered domain name.

Every node stores a hash table in addition to storing its own data. This is used to record file storage locations to allow file querying and downloads.

Our previous article mentioned the shortcomings of HTTP and centralized servers; so the question becomes, how does IPFS solve these shortcomings?

The download speed is fast, because it doesn’t rely on network backbones or centralized servers.

The entire IPFS system is a distributed file storage system, so it downloads information from several nodes when accessing relevant data. This is much faster than HTTP’s centralized server speed. Everyone’s used P2P to download files before (eg Xunlei or BitTorrent); IPFS’ download method is similar to this.

Storage Space Becomes Extremely Cheap

Because IPFS uses blockchain technology, it utilizes Filecoin (the article will describe how to obtain Filecoin, that is, mining) to reward miners for sharing their hard drive. IPFS also removes redundant storage from the network (this greatly saves storage space across the entire network). IPFS storage space will be ridiculously cheap in the future (when compared to our current cloud and central CND storage).

Security

It’s very difficult for modern centralized servers to combat DDoS attacks. Centralized servers practically become paralyzed when a large number of requests come from multiple locations at once. Any who have studied operations ought to understand this problem. Like during the yearly 11/11 sale, our friends at Alibaba, Tengxun technology, and the entire banking industry have to work overtime. A massive number of requests can instantly force a server into downtime. IPFS is able to resist this type of attack due to its nature. This is because all requests are distributed across a variety of nodes, so that even the attacker becomes their own node. To a certain extent, IPFS can even resist quantum computing attacks.

Open

Everyone knows, Bitcoin is a well-known centralized currency. These characteristics make Bitcoin impossible to regulate, and transactions impossible to forge. In the same vein, due to IPFS being built on a decentralized distributed network, it’s difficult for IPFS to centrally controlled or limited, making the internet more open.