Distributed File System
Subject: Computer and Technology noun /dɪˈstrɪb.jə.tɪd faɪl ˈsɪs.təm/ 1 definition
- distributed-systems
- file-storage
- storage-architecture
- operating-systems
-
Definition (noun)
Entry #128 · UpdatedA Distributed File System (DFS) is a storage architecture that stores and manages files, directories, and metadata across multiple networked servers while exposing them to clients through a single unified namespace. By decoupling logical file paths from physical storage locations, a DFS coordinates data placement, caching, access control, and locking across cluster nodes so applications can read and write shared files without tracking which machine holds each block. Built-in replication or erasure coding delivers horizontal scalability and fault tolerance when individual drives or servers fail, distinguishing a DFS from single-host local file systems and flat object stores while introducing network latency and distributed consistency trade-offs.
Common Types
Client-Server File Systems: Mount exported directory trees from storage servers onto remote clients over protocols such as NFS or SMB, preserving familiar POSIX semantics for shared enterprise volumes and home directories.
Cluster and Parallel File Systems: Stripe file data across multiple storage servers simultaneously—as in Lustre, IBM Spectrum Scale, or CephFS—to deliver high aggregate I/O throughput for high-performance computing and clustered workloads.
Chunked Big-Data File Systems: Partition large files into fixed multi-megabyte blocks replicated across commodity nodes under centralized metadata coordinators, such as Google File System (GFS) and Hadoop Distributed File System (HDFS), optimizing for sequential batch reads and append-heavy analytics.
Peer-to-Peer Decentralized File Systems: Distribute file chunks and lookup metadata across equal nodes without a central coordinator—often using distributed hash tables or content addressing as in IPFS—to support wide-area resilience without single points of failure.
Example
A genomics lab stores 120 terabytes of sequencing data on a 24-node CephFS cluster so compute workers on separate machines can simultaneously read
/shared/runs/sample_042.fastqthrough a standard directory mount. When storage node 11 goes offline for a disk replacement, the distributed file system transparently serves replicated data chunks from healthy peer nodes so running analysis jobs finish without failing.