Abstract: An amount of expiration time extension for one or more objects associated with a first archive of a first snapshot of a source storage is determined based at least in part on a second data management policy associated with a second archive and one or more dynamically determined metrics. The first archive that includes the one or more objects is caused to be stored to a remote storage. At least a portion of content of the first archive is referenced by data chunks stored in a first chunk object of the remote storage and the first archive is associated with a first data management policy. Based on the determined amount of expiration time extension, an expiration time for the one or more objects associated with the first archive is stored in an archive metadata of the one or more objects associated with the first archive.

Abstract: File metadata structures of a file system are analyzed. At least one metadata element that is duplicated among the analyzed file metadata structures is identified. The at least one identified metadata element is deduplicated including by modifying at least one of the file metadata structures to reference a same instance of the identified metadata element that is referenced by another one of the file metadata structures.

Abstract: Techniques are described for efficient and secure key rotation for backup storage. An example method comprises generating, by a data platform implemented by a computing system, a first encrypted key and a second encrypted key, wherein the data platform stores one or more encrypted chunks encrypted using a first encryption key of the first encrypted key and a second encryption key from the second encrypted key, the first encrypted key and the second encrypted key encrypted with a first system key, replacing the first system key by determining a second system key, and encrypting, with the second system key, the first encryption key to generate a replacement first encrypted key and the second encryption key to generate a replacement second encrypted key.

Abstract: Techniques are described for identifying deleted keys during snapshot comparisons managed by a data platform. A computing system that implements the data platform comprising a memory and processing circuitry may be configured to perform the techniques. The memory may store metadata identifying, at least in part, one or more deleted keys representative of deleted data that has been deleted from a storage system. The processing circuitry may identify a first tree data structure corresponding to a first snapshot of the storage system at a first time, and identify a second tree data structure corresponding to a second snapshot of the storage system at a second later time. The processing circuitry may traverse the first tree data structure and second tree data structures to identify the one or more deleted keys, and perform, based on the deleted keys, a data protection operation.

Abstract: A backup of one or more objects is determined to be performed. Based on one or more conditions, a corresponding deduplication option among a plurality of deduplication options to utilize when backing up the one or more objects is selected. The one or more conditions at least include a condition based on a detected data change pattern. The plurality of deduplication options include a deduplication option associated with utilizing at least in part a plurality of variable-length data chunks for one or more mismatched ranges and/or one or more missing ranges associated with one of the one or more objects associated with the source system. A request to perform the backup of the one or more objects according to the corresponding selected deduplication option is provided to the source system. Backup data associated with the one or more objects is received and stored.

Abstract: Techniques are disclosed for adaptive throttling based on rate-limiting signals from cloud infrastructure. An example method comprises receiving, by a data platform implemented by a plurality of nodes, a service identifier from a cloud infrastructure system, receiving, by a first node of the plurality of nodes, a rate-limiting signal from the cloud infrastructure system, updating, by the data platform, context information associated with the service identifier based on the rate-limiting signal, sending, by a second node of the plurality of nodes and based on the service identifier, a request for the cloud infrastructure system to the first node, and throttling, by the first node, the request for the cloud infrastructure system based on the context information at least for a delay period.

Abstract: One or more event logs are received. The one or more event logs are analyzed using a plurality of models to detect one or more anomalous events. A graphical representation of risk entities associated with at least one of the one or more detected anomalous events is provided. A visual representation of automatically detected relationships between the risk entities associated with the at least one of the one or more detected anomalous events is provided in the graphical representation. Indications of measures of anomaly associated with detected anomalous events are provided for the associated risk entities.

Abstract: An archival storage of data backed up from a repository storage of a primary storage is maintained. Access to data stored in archival storage is limited by one or more access policies based on whether a corresponding data restore has been authorized. A request for specific data stored in the archival storage is received. The one or more access policies are automatically managed based on status and timing of one or more data restore authorizations for the specific data stored in the archival storage.

Abstract: Techniques are described for data management across cloud environments. An example method comprises restoring, by the data platform, a selected copy of the one or more copies from the one or more second storage systems by, for each chunk of the selected copy: identifying, by the data platform, the chunk in the chunk metadata, determining, by the data platform, whether a matching chunk is stored on the first storage system based on the chunk metadata, responsive to determining the matching chunk is stored on the first storage system, retrieving, by the data platform, the matching chunk from the first storage system, wherein the matching chunk is included in the selected copy, and responsive to determining the matching chunk is not stored on the first storage system, retrieving, by the data platform, the chunk from the one or more second storage systems, wherein the chunk is included in the selected copy.

Abstract: In general, techniques for efficient data retrieval from a backup system are described. An example computing system includes one or more storage devices and processing circuitry having access to the one or more storage devices and configured to: process an input to generate a filter, wherein the input indicates a context for one or more queries; apply the filter to backup data to obtain filtered data from the backup data; generate an index of embeddings from the filtered data; process, based on the index of embeddings, a query to generate a response for the query; and output the response.

Abstract: Range information associated with one or more objects is received from a storage system. One or more missing ranges and/or one or more mismatched ranges associated with the one or more objects is determined based on the received range information. A plurality of data chunk identifiers associated with a plurality of variable-length data chunks included in the one or more determined ranges associated with the one or more objects is provided to the storage system. A response that is used to identify among the plurality of variable-length data chunks, one or more variable-length data chunks not already stored in a storage associated with the storage system is received from the storage system. Content of the identified one or more variable-length data chunks is provided to the storage system.

Abstract: Techniques are disclosed for prefetching data using predictive analysis. An example method comprises storing, by a data platform implemented by a computing system, objects of a file system, wherein a first subset of the objects is stored to a first storage tier and a second subset of the objects is stored to a second storage tier, classifying objects into one or more classifications, storing a data access record for the objects, applying a machine learning model to generate a prediction of future data access to one or more objects of the second subset based on the one or more classifications and the data access record, wherein the prediction includes a predicted time for the future data access, and retrieving, based on the prediction, the one or more objects of the second subset from the second storage tier prior to the predicted time.

Abstract: Methods, computer program products, computer systems, and the like for improved performance, when verifying backup images, are disclosed, which can include generating representational information (where the representational information is based, at least in part, on at least a portion of information of a top-level entity), retrieving a stored representational information (where the stored representational information is based, at least in part, on the at least the portion of the information of the top-level entity, the stored representational information is retrieved from log information produced as part of a backup operation, and the backup operation produced a backup image of the top-level entity), determining whether the representational information and the stored representational information match, and, in response to a determination that the representational information and the stored representational information do not match, indicating that the backup image of the top-level entity is corrupt.

Abstract: An assigned subgroup that includes a plurality of entries is traversed by a prefetcher. It is determined that an expected number of entries associated with the assigned subgroup have been traversed. In response to determining that expected number of entries associated with the assigned subgroup have been traversed, it is determined that a last read entry associated with the assigned subgroup does not correspond to a last entry associated with the assigned subgroup. The prefetcher is preempted by stopping the prefetcher from obtaining a list of entries associated with a remaining portion of the assigned subgroup.

Abstract: A plurality of different views of data associated with a storage domain stored on a deduplicated storage are traversed to determine data chunks belonging to each view of the plurality of different views of data associated with the storage domain. A request for a metric associated with disk space utilization of a group of one or more selected views of data associated with the first storage domain included in the plurality of different views of data associated with the first storage domain that are stored on the deduplicated storage is received. Data chunks belonging to the one or more selected views of data associated with the first storage domain of the group are identified. An incremental disk space utilization of the group is determined, including by determining a total size of the identified data chunks. The metric associated with disk space utilization is provided based on the determined incremental disk space utilization of the group.

Abstract: An inode is associated with an incomplete file system operation to a file system object is determined based on an update intent stored in the inode. It is determined that a set of inodes is associated with the incomplete file system operation based on the update intent stored in the inode. The update intent indicates the set of inodes is associated with the incomplete file system operation. It is determined to complete the incomplete file system operation based on evaluating the set of inodes that is associated with the incomplete file system operation. Evaluating the set of inodes that is associated with the incomplete file system operation includes determining whether the set of inodes that is associated with the incomplete file system operation stores a corresponding update intent. The incomplete file system operation is completed based on the evaluation of the set of inodes that is associated with the incomplete file system operation.

Abstract: It is determined that an expiration date for an object associated with a first archive that includes a chunk object that includes a plurality of data chunks has expired. The first archive is stored in a first cloud performance storage class according to an archive tiering policy. It is determined that the archive tiering policy indicates migrating the plurality of data chunks included in the chunk object from the first cloud performance storage class to a second cloud performance storage class. In response to determining that the archive tiering policy indicates migrating the plurality of data chunks included in the chunk object from the first cloud performance storage class to the second cloud performance storage class, the plurality of data chunks included in the chunk object are migrated from the first cloud performance storage class to the second cloud performance storage class.

Abstract: Techniques are disclosed for efficient file system directory reverse name lookup. An example method includes storing, at one or more of a plurality of storage locations of a key map, a file system directory including one or more key-value pairs, wherein each key-value pair of the file system directory maps a respective object identifier to corresponding object metadata; and in response to detecting an indication of a file system event for a file: storing, to a storage location of the key map, a key-value pair that maps an object identifier of the file to object metadata of the file; generating, based on the object identifier, an indication of the storage location; including, by the data platform, the indication of the storage location in the object metadata; and storing, by the data platform, the object metadata to a data store.

Abstract: Techniques are described for fast large file truncation. An example method includes receiving, by a data platform, a request to truncate a file, wherein data of the file is stored in a tree data structure including a plurality of nodes with leaf nodes corresponding to data of the file; determining a first node of the plurality of nodes including a plurality of child nodes; determining, based on a maximum key, a subset of the child nodes corresponding to a portion of the data that is to be retained; based on determining not to traverse the first node, updating a second node that is a parent of the first node to include a pointer to each child node of the subset of child nodes; and moving, from a subset of leaf nodes referenced by the subset of child nodes, a leaf node to rebalance the tree data structure.

Abstract: A repository of replicated chunk files is analyzed to identify chunk files that meet at least a portion of combination criteria. Selected chunk files are associated together under a data protection grouping container. Erasure coding is applied to the data protection grouping container including by utilizing the selected chunk files as different data stripes of the erasure coding and generating one or more parity stripes based on the different data stripes.