A scalable de-duplication file system divides the file system into data and metadata stores where each store is built on scale out architectures. Each store is not a single module, but a collection of identical modules that together creates one large store. By scaling metadata store and chunk store, the file system can be scaled linearly without compromising the file system performance. Deduplication logic identifies a chunk location for each stored chunk, and stores, for each identifier, an index of the chunk location associated with the corresponding identifier, such that the stored index for similar chunk ids points to the same chunk location. In this manner, duplicate chunks or blocks of data are referenced merely by pointer or indexes, rather than redundantly duplicating storage for each instantiation or copy of similar data.