Abstract: Corruption detection in backups is disclosed. Backups that are received into a backup environment are stored in corresponding lineages. A detection engine is configured to perform corruption detection operations on the most recent backups in each of the lineages based on a sample frequency. Corruption detection operations may also be performed randomly and based on unexpected or unusual changes in backup metadata.

Abstract: Improving the performance of read operations in a restore path of an inline deduplication system utilizing a DDBOOST interface by providing an adaptive compression component for use with DDBOOST applications. A built-in compression mode transfers read data if there are sufficient CPU resources in the server and client to compress and decompress the read data without destabilizing the system. CPU usage is tracked to generate predicted respective client and server CPU usage. These respective predictions are compared to defined maximum threshold usage values. If the predicted values do not exceed the thresholds, compression is used, otherwise the data is transmitted over the network as non-compressed data. A pre-filter is used to first determine whether or not the data would benefit from the built-in compression mode.

Abstract: A self-healing method of recovering files using a cyber recovery vault for deduplication filesystems. Data to be backed up from a production system to a backup system in a datacenter is also copied to a cyber security vault coupled to the data center through an air gapped connection. A copy process makes a point-in-time (PIT) copy of the portion of data to a namespace in the vault, and retention locks the copied portion in the vault. An automated process detects a missing segment error, such as due to a bad index lookup in the data to be backed up. The self-healing process provides the missing segment from the PIT copy in the cyber security vault. For any missing fingerprints the process adds the data in new containers and the references in the index, providing an indirection that allows for just writing a new container with the missing L0 references and updating the index.

Abstract: Methods of cloning data backup across namespaces are disclosed. In response to receiving a file replication request, one or more source files are cloned from a first namespace to a second namespace, as one or more destination files. A first attribute indicating that the source file(s) have been cloned to the second namespace as the destination file(s) is created. A source synthetic file is cloned from the first namespace to the second namespace, as a destination synthetic file, where synthetic information of the destination synthetic file points to the source file(s) as one or more base file. A second attribute indicating that the source synthetic file has been cloned to the second namespace as the destination synthetic file is created. Based on the first and second attributes, the synthetic information of the destination synthetic file is updated or propagated to use the destination file(s) as one or more base files on the second namespace.

Abstract: Improving the performance of read operations in a restore path of an inline deduplication system utilizing a DDBOOST interface by providing an adaptive compression component for use with DDBOOST applications. A built-in compression mode transfers read data if there are sufficient CPU resources in the server and client to compress and decompress the read data without destabilizing the system. CPU usage is tracked to generate predicted respective client and server CPU usage. These respective predictions are compared to defined maximum threshold usage values. If the predicted values do not exceed the thresholds, compression is used, otherwise the data is transmitted over the network as non-compressed data. A pre-filter is used to first determine whether or not the data would benefit from the built-in compression mode.

Abstract: Embodiments for providing file immutability for cloud storage data in a deduplicating filesystem and using a new filesystem that is spawned to receive redirected live data, after which the old filesystem is expired. Data objects are stored in the cloud by defining a protection duration from a first date to a fixed future date, and applying a retention lock to one or more data objects stored in cloud storage during the protection duration, the retention lock preventing unauthorized deletion, modification or movement of the data. A renew threshold date is defined within the protection duration, and a new filesystem is spawned on this date. All new data is redirected to the new filesystem and locked for the protection duration, and the first file system is destroyed.

Abstract: Embodiments for processing fastcopy overwrite backup workloads (FCOW) as virtual synthetic backups, by detecting whether or not a backup workload comprises an FCOW workload in which a base file is fastcopied to a target file, wherein portions of the base file are overwritten through overwrites by new data in the target file by a deduplication backup process. For FCOW workloads, the process creates a file recipe by trapping seek offsets and write byte lengths for the overwrites, and writing extent information comprising offset differences to a recipe table of the file recipe, and the file recipe facilitates file processing optimizations including differencing, replication, and tiering.

Abstract: Corruption detection in backups is disclosed. Backups that are received into a backup environment are stored in corresponding lineages. A detection engine is configured to perform corruption detection operations on the most recent backups in each of the lineages based on a sample frequency. Corruption detection operations may also be performed randomly and based on unexpected or unusual changes in backup metadata.

Abstract: Embodiments for performing the inline deduplication by filtering streaming data as it is received by a backup client through a backup server executing a backup process. A data processing unit (DPU) is deployed to offload certain processing operations performed by a central processing unit (CPU) of the backup client. An inline deduplication operation comprises file operations, data segmentation, segment fingerprinting, compression, and encryption prior to storage in a backup target. The DPU is deployed and configured to perform the compression and encryption steps. It is also configured to perform decompression tasks offloaded by the client for reading compressed data during restore operations.

Abstract: Improving data recovery and restoration times by clamping down security after malware detection, and protect a data center from a cyber recovery vault. Embodiments provide a cyber recovery vault configured to store data backed up for a production site for long-term retention and disaster recovery. The vault is coupled to the data center comprising a production site through an automated air gap controlled by the vault. Control signals transmitted by the vault trigger the air gap to close the coupling between the vault and data center upon detection of a malware attack, and the data center is configured to listen for the control signals and implement heightened security measures to protect its data in response to the control signal.

Abstract: Embodiments for performing the inline deduplication by filtering streaming data as it is received by a backup client through a backup server executing a backup process. A data processing unit (DPU) is deployed to offload certain processing operations performed by a central processing unit (CPU) of the backup client. An inline deduplication operation comprises file operations, data segmentation, segment fingerprinting, compression, and encryption prior to storage in a backup target. The DPU is deployed and configured to perform the compression and encryption steps, the entire inline deduplication stack, or the entire inline deduplication stack plus the file system operations.

Abstract: Methods of cloning data backup across namespaces are disclosed. One or more source files are cloned from a first namespace to a second namespace, as one or more destination files. When the cloning of the source file(s) is performed, a data structure including source file information and destination file information is generated. A source synthetic file is cloned from the first namespace to the second namespace, as a destination synthetic file, where the source synthetic file uses the source file(s) as one or more base files on the first namespace. When the cloning of the source synthetic file is performed, the data structure is looked up to obtain the source file information and the destination file information. Based on the source file information and the destination file information, synthetic information of the destination synthetic file is updated to use the destination file(s) as one or more base files on the second namespace.

Abstract: This disclosure provides system, methods, and media for identifying inadvertent compression or encryption in data streams from a client that land on a deduplication storage system. When one or more such abnormalities are detected, an alert message is generated to alert the administrator of the deduplication storage system so that corrective actions can be taken to prevent undesired consequences. According to an exemplary method, machine learning techniques are used to plot and smoothen global compression ratios and local compression ratios of historical backups from a client over a period of time. Then, a second derivative of each data point on the smoothened curves is taken and compared with a predetermined threshold to detect whether that the slope of the data point exceeds a threshold. A data point whose slope exceeds the threshold can be determined to be a data point corresponding to a backup that includes compression and/encryption.

Abstract: Improving the performance of read operations in a restore path of an inline deduplication system utilizing a DDBOOST interface by providing an adaptive compression component for use with DDBOOST applications. The system utilizes a built-in compression mode for transferring read data if there is a sufficient available CPU resources in both the server and client to respectively compress and decompress the read data without destabilizing the system. CPU usage on both the client and the server is tracked to generate predicted respective CPU usage. These respective predictions are compared to defined maximum threshold usage values. If the predicted values do not exceed the thresholds, compression is used, otherwise the data is transmitted over the network as non-compressed data.

Abstract: Described is a system for leveraging incremental backup metadata of a deduplication storage system to perform an efficient data restore. The system may reference metadata stored as part of incremental backup infrastructure to determine only the changes to the data that occurred subsequent to the restore point. As a result, the system may reduce data read processing by retrieving only the changed data from the deduplication storage system. Moreover, the system may reduce the required bandwidth for the restore by providing only the changed data to a target device of the restore. Data retained by the client system such as the current version of the data may then be overwritten with the reduced set of data to complete the restore. As a result, the system may perform an efficient restore by reducing read accesses and bandwidth requirements, and thus, improve overall restore performance.

Abstract: Corruption detection in backups is disclosed. Backups that are received into a backup environment are stored in corresponding lineages. A detection engine is configured to perform corruption detection operations on the most recent backups in each of the lineages based on a sample frequency. Corruption detection operations may also be performed randomly and based on unexpected or unusual changes in backup metadata.

Abstract: Embodiments for providing file immutability for cloud storage data in a deduplicating filesystem and using a new filesystem that is spawned to receive redirected live data, after which the old filesystem is expired. Data objects are stored in the cloud by defining a protection duration from a first date to a fixed future date, and applying a retention lock to one or more data objects stored in cloud storage during the protection duration, the retention lock preventing unauthorized deletion, modification or movement of the data. A renew threshold date is defined within the protection duration, and a new filesystem is spawned on this date. All new data is redirected to the new filesystem and locked for the protection duration, and the first file system is destroyed.

Abstract: Embodiments for providing file immutability for cloud storage data in a deduplicating filesystem and using a new filesystem that is spawned to receive redirected live data, after which the old filesystem is expired. Data objects are stored in the cloud by defining a protection duration from a first date to a fixed future date and a selection period to select a subset of data objects to be protected during the protection period. A retention lock is applied to files that meet a set selection criteria, as determined by a matching or filter process. The selection criteria can include one or more of time-based selection, filetype selection, or tag-based selection.

Abstract: Embodiments for providing file immutability for cloud storage data in a deduplicating filesystem and using a new filesystem that is spawned to receive redirected live data, after which the old filesystem is expired. Data objects are stored in the cloud by defining a protection duration from a first date to a fixed future date and a selection period to select a subset of data objects to be protected during the protection period. A retention lock is applied to the subset of data objects stored in cloud storage during the protection duration, the retention lock preventing unauthorized deletion, modification or movement of the data. A cloud bucket is created for storing the subset of data objects and a new filesystem is spawned for attachment to the cloud bucket. Upon expiration of the selection period, the retention locks are expired and the new filesystem is destroyed.

Abstract: A method for providing data immutability using a deduplication filesystem for data objects stored in the cloud by defining a protection duration from a first date to a fixed future date, and applying a retention lock to one or more data objects stored in cloud storage during the protection duration, the retention lock preventing deletion, modification or movement of the data objects by an unauthorized entity. The method defines a renew threshold date within the protection duration, and performs a garbage collection (GC) operation at the renew threshold date to permanently delete dead data objects and carry forward live data objects, and extends the retention lock on remaining live data objects and subsequently written data objects upon completion of the GC operation.