Skip to content

IPFS.md

Latest commit

 

History

History
81 lines (58 loc) · 3.6 KB

IPFS.md

File metadata and controls

81 lines (58 loc) · 3.6 KB
Raw

IPFS

Content Addressing

What is a CID?

  • A particular form of Content Addressing for the Decentralized Web.
  • It was developed for IPFS, but has broad implications.
  • It contains Cryptographic Hash and a Codec.
  • Codecs encode and decode data in certain formats.
  • Git, Bitcoin and Ethereum all use content addressing.
  • CID creates an universal identifier for any of these systems.
+-------+------------------------------+
| Codec | Multihash                    |
+-------+------------------------------+
  • Multihash or the self-describing hash, contains Hash Type and Hash Value.
+------------------------------+
| Codec                        |
+------------------------------+
|                              |
| Multihash                    |
| +----------+---------------+ |
| |Hash Type | Hash Value    | |
| +----------+---------------+ |
|                              |
+------------------------------+
  • We generate fingerprint by using a cryptogrphic algorithm that maps input of arbitrary size (data or a file) to output of fixed size.
  • We call the output cryptographic hash digest or simply hash.

Fingerprint

The cryptographic algorithm used must generate hashes that have the following characteristics:

  • Deterministic: The same input should always produce the same hash.
  • Uncorrelated: A small change in the input should generate a completely different hash.
  • One-way: It should be infeasible to reconstruct the data from the hash.
  • Unique: Only one file can produce one specific hash.

IPFS uses sha2-256 by default, though a CID supports any strong cryptographic hash algorithm.

Multihash

  • Hashing algorithms may prove to be insecure through time, so multihash adds the ability to know which algorithms was used to generate the hash, thus making the CID generalized.
  • For this reason it is called the self-describing hash.
  • Multihash follow the TLV pattern (type-length-value).

Multihash Format

  • type: identifier of the cryptographic algorithm used to generate the hash (e.g. the identifier of sha2-256 would be 18 - 0x12 in hexadecimal) - see the multicodec table for all the identifiers

  • length: the actual length of the hash (using sha2-256 it would be 256 bits, which equates to 32 bytes)

  • value: the actual hash value

  • To display 0 and 1 of the output to string, IPFS uses base58btc encoding. This version was known as CID0.

  • The initials Qm... is thus common in IPFS.

Problems with CID0

  • How do we know what method was used to encode the data?
  • How do we know what method was used to create the string representation of the CID? Will we always be using base58btc?

These problems were addressed in CID1, which is the current version of CID.

CID1 : Multicodec Prefix

  • This new version adds a new prefix known as Multicodec Prefix which tells which encoding was used on the data.
  • It could tell whether it is encoded with CBOR, Protobuf, plain JSON, etc.
  • Here's the complete table.

CID01

  • The dag-pb is one of many different types of IPLD (InterPlanetary Linked Data) codecs.
  • This multicodec is not only limited to IPFS and IPLD but is a part of the Multformats project.
  • Multiformats project has implications in wide variety of projects including the CID.