Abstract: A method comprising: stopping to use one or more first approximate membership filters for servicing of read requests that are incoming to a storage system; increasing a size of a set of first data structures that are available in the storage system, each of the first data structures being used by the storage system for write request amortization; replacing the one or more first approximate membership filters with one or more second approximate membership filters, and building each of the one or more second approximate membership filters by: freezing all non-empty first data structures in the set, and copying respective contents of each of the frozen non-empty first data structures into any of the one or more second approximate membership filters; and starting to use the one or more second approximate membership filters for servicing of read requests that are incoming to the storage system.

Abstract: A technique for managing metadata of variable length includes responding to the creation or change in a metadata element by creating at least first and second entries within a metadata page at discontinuous locations. The first entry is located among a first set of regions having uniform length and includes a reference to the second entry, which is located among a second set of regions having variable length. In this manner, the metadata element that does not fit within a fixed-size space is accommodated by multiple discontinuous entries in respective sets of regions.

Abstract: Data units of a dataset may be compressed by clustering the data units into clusters, selecting a reference unit for each unit cluster, and compressing data units of each unit cluster using the reference unit of the unit cluster as a dictionary. The computational efficiency of the clustering algorithm may be improved by not applying it to data units themselves, but rather to hash values of the data units, where the hash values have a much smaller size than the data units. The hash function may be a locality-sensitive hash (LSH) function. The reference unit of a cluster may be determined in any of a variety of ways, for example, by selecting a centroid or exemplar of the cluster. Clusters, including their references values, may be indexed in a cluster index (e.g., a Faiss index), which may be searched to assign future added or modified data units to clusters.

Abstract: A method comprising: stopping to use one or more first approximate membership filters for servicing of read requests that are incoming to a storage system; increasing a size of a set of first data structures that are available in the storage system, each of the first data structures being used by the storage system for write request amortization; replacing the one or more first approximate membership filters with one or more second approximate membership filters, and building each of the one or more second approximate membership filters by: freezing all non-empty first data structures in the set, and copying respective contents of each of the frozen non-empty first data structures into any of the one or more second approximate membership filters; and starting to use the one or more second approximate membership filters for servicing of read requests that are incoming to the storage system.

Abstract: Data units of a dataset may be compressed by clustering the data units into clusters, selecting a reference unit for each unit cluster, and compressing data units of each unit cluster using the reference unit of the unit cluster as a dictionary. The computational efficiency of the clustering algorithm may be improved by not applying it to data units themselves, but rather to hash values of the data units, where the hash values have a much smaller size than the data units. The hash function may be a locality-sensitive hash (LSH) function. The reference unit of a cluster may be determined in any of a variety of ways, for example, by selecting a centroid or exemplar of the cluster. Clusters, including their references values, may be indexed in a cluster index (e.g., a Faiss index), which may be searched to assign future added or modified data units to clusters.