Technical Glossary
A peer-to-peer distributed file system that uses content-addressing to uniquely identify and retrieve data across a decentralized network of nodes. IPFS replaces location-based addressing with cryptographic hashes derived from file content, ensuring data integrity and enabling deduplication across the network. The protocol combines ideas from distributed hash tables, BitTorrent-style file distribution, and Git-like versioned data structures. IPFS is maintained by Protocol Labs and has become foundational infrastructure for Web3 applications, NFT metadata storage, and censorship-resistant publishing.
A self-describing content-addressed identifier that uniquely labels a piece of data based on its cryptographic hash, ensuring that identical content always produces the same identifier regardless of where or when it is stored. CIDs encode the hash function used, the codec for data serialization, and the multihash digest into a compact, version-prefixed string. Version 1 CIDs support multiple hash algorithms and codecs through the multiformat self-describing framework. CIDs are the fundamental addressing primitive in IPFS and related content-addressed systems like Filecoin and IPLD.
A decentralized key-value lookup system that distributes data location information across participating nodes using consistent hashing algorithms to enable efficient content discovery without central coordination. DHTs in IPFS implement the Kademlia protocol, organizing nodes into a structured overlay network where each node maintains routing information for logarithmically scaling peer discovery. The DHT enables nodes to locate content providers by querying progressively closer nodes to the target hash. Performance optimizations include accelerated DHT mode, provider record republication, and routing table maintenance protocols.
An HTTP-based access point that bridges traditional web browsers to the IPFS network by fetching content-addressed data and serving it through standard HTTP responses. Gateways enable users without IPFS node software to access distributed content using familiar URL patterns that translate CIDs into HTTP requests. Public gateways provide universal read access while private or dedicated gateways offer performance optimization, access control, and custom domain configuration. The IPFS Gateway specification defines path resolution, subdomain isolation, and response header requirements for compliant gateway implementations.
A service that ensures specific content remains persistently available on the IPFS network by maintaining pinned copies on dedicated infrastructure nodes that exempt the data from garbage collection. Pinning addresses the challenge of content availability in IPFS, where unpinned data may be removed during routine cache maintenance on individual nodes. Remote pinning services provide API-driven content persistence without requiring users to operate their own infrastructure. The IPFS Pinning Service API specification standardizes the interface for managing remote pin operations across different service providers.
A data model specification that enables the creation of hash-linked data structures traversable across protocol boundaries using content-addressed identifiers as universal links. IPLD defines a canonical data model based on a type system of scalars, lists, maps, links, and bytes that can be serialized into multiple codecs including DAG-CBOR and DAG-JSON. The framework enables applications to navigate complex data graphs spanning IPFS, Ethereum, Git, and other content-addressed systems through a unified path-based traversal mechanism. IPLD schemas provide structural type definitions that enable validation and code generation for typed data handling.
A directed acyclic graph data structure where each node is identified by the cryptographic hash of its content and links, enabling verifiable data integrity throughout the entire graph structure. Merkle DAGs form the underlying storage model in IPFS, representing files, directories, and arbitrary data as interconnected content-addressed nodes. Any modification to a leaf node produces cascading hash changes up through the graph, providing tamper-evident properties without requiring a trusted authority. The structure enables efficient synchronization, deduplication, and incremental verification of large datasets across distributed systems.
The data exchange protocol used by IPFS nodes to request, send, and receive content blocks from peers in the network through a message-based want-list and block trading mechanism. Bitswap maintains ledgers tracking block exchanges between peers, implementing a credit-based incentive system that rewards nodes contributing content to the network. The protocol optimizes bandwidth utilization through want-have and want-block message types that enable nodes to discover block availability before requesting transfers. Bitswap operates at the block level, collaborating with the DHT and content routing layer to locate peers holding requested content.
A mutable naming layer built on top of IPFS that enables cryptographic key holders to publish signed pointers from persistent names to changing content identifiers, providing stable addresses for dynamic content. IPNS records are signed by the key owner and published to the DHT or other routing systems, allowing anyone to verify the authenticity of the name-to-CID mapping. The system addresses the inherent immutability of content addressing by decoupling human-meaningful or persistent names from the underlying content hashes. IPNS records include sequence numbers and expiration timestamps to support ordered updates and prevent replay attacks.
A decentralized storage network built on IPFS infrastructure that uses blockchain-based incentives and cryptographic proofs to create a verifiable marketplace for persistent data storage. Filecoin storage providers commit disk space and prove ongoing data custody through Proof-of-Replication and Proof-of-Spacetime consensus mechanisms. Clients negotiate storage deals specifying duration, redundancy, and price parameters through an on-chain marketplace protocol. The network provides economic guarantees for long-term data availability that complement IPFS content distribution with incentivized persistence.
The subsystem responsible for discovering which network nodes hold specific content blocks by querying distributed routing infrastructure to map content identifiers to provider addresses. Content routing in IPFS operates through multiple mechanisms including the Kademlia DHT, delegated routing endpoints, and network indexer services that aggregate provider records. The routing layer must balance discovery latency, bandwidth overhead, and privacy considerations when locating content across the global peer network. Delegated content routing through HTTP APIs enables lightweight clients to discover content without participating in the full DHT protocol.
A data format specification used by IPFS to represent files, directories, and symbolic links as Merkle DAG structures with metadata including file sizes, chunk boundaries, and directory listings. UnixFS implements file chunking algorithms that split large files into fixed-size or content-defined blocks, each independently content-addressed and distributable across the network. The format preserves enough filesystem semantics to support directory traversal, file assembly, and size reporting while remaining compatible with content-addressed storage. UnixFS v2 proposals aim to improve efficiency and metadata extensibility for large-scale file handling.
A modular peer-to-peer networking framework originally developed for IPFS that provides transport-agnostic connectivity, peer discovery, stream multiplexing, and secure communication primitives for distributed applications. Libp2p abstracts underlying transport protocols including TCP, QUIC, and WebSocket behind a unified interface, enabling applications to communicate across heterogeneous network environments. The framework includes NAT traversal through relay protocols, authenticated connections via Noise or TLS handshakes, and publish-subscribe messaging through GossipSub. Libp2p has been adopted beyond IPFS by projects including Ethereum, Polkadot, and Filecoin for their peer-to-peer networking requirements.
A deterministic serialization format based on Concise Binary Object Representation that provides the canonical encoding for structured data within the IPLD ecosystem and IPFS storage layer. DAG-CBOR restricts the full CBOR specification to ensure consistent encoding, requiring sorted map keys, specific integer encoding rules, and link representation through CID tag 42. The format enables compact binary representation of complex data structures while maintaining content-addressable properties through deterministic serialization. DAG-CBOR is specified as an IPLD codec and leverages the IETF CBOR standard with additional constraints for hash-consistent encoding.
The storage management process in IPFS nodes that reclaims disk space by removing cached content blocks that are not pinned or referenced by any active data structures. Garbage collection traverses the node's block store, identifies unreferenced blocks, and deletes them to prevent unbounded storage growth from transient content retrieval. Pinned content is protected from garbage collection, ensuring that intentionally preserved data remains available regardless of cache pressure. Node operators configure garbage collection thresholds, triggering policies, and storage limits based on their available disk capacity and content hosting commitments.