Abstract: Techniques are provided involving combining several concurrent changes into one consistent condensed transaction within a metadata transaction log using an accumulation list that can be processed to generate a single transaction which may be quickly stored in the metadata transaction log. The improved techniques thus reduce latencies for later concurrent changes and save space within the metadata transaction log.

Abstract: A technique performs a host write operation on behalf of a host computer. The technique involves receiving a write request from the host computer, and providing a data write command and a zero write command to backend storage in response to the write request from the host computer. The backend storage maintains a file on behalf of the host computer. Additionally, the data write command includes host data to be written to the file. Furthermore, the zero write command defines a field of zeroes to be written to the file. The technique further involves sending an acknowledgement to the host computer, the acknowledgement serving as a response to the write request from the host computer.

Abstract: Embodiments of the present disclosure relate to a method and apparatus for accessing data by receiving a data read request for reading data corresponding to a logical block number; determining a first physical block corresponding to the first logical block number on the disk when a first cache page corresponding to the first logical block number does not exist in a cache; and reading data in the second cache when a second cache corresponding to the second physical block exists in the cache, wherein the content of the second physical block is identical to the content of the first physical block. Some embodiments of the present disclosure may prevent storing mass redundant data, and thereby enhance data reading rate.

Abstract: The present disclosure relates to a method and apparatus for a file system, the file system comprising at least a first logical storage area and a second logical storage area, a logical block of the first logical storage area and that of the second logical storage area being different in size, and in response to a request for creating a file, selecting a logical storage area for storing the file from the at least first logical storage area and second logical storage area; and storing the file in the selected logical storage area. According to the above various embodiments of the present invention, a direct and effective way may be used to support variable-length block size in a file system, without increasing metadata and positioning overheads.

Abstract: A technique for managing usage metrics of quota trees in a file system includes storing, in an inode table that lists inodes of the file system, identifiers (IDs) of quota trees to which the inodes belong. To check the usage of a quota tree, the inode table is scanned for any inodes that reference the quota tree ID of that quota tree. All such inodes are identified, and their usage attributes are summed to calculate an overall usage metric for the quota tree as a whole. The improved technique thus avoids the need to scan directory structures on disk.

Abstract: Embodiments of the present disclosure provide a data transfer method and apparatus by establishing, at a source device, a plurality of data connections with a target device in response to a data connection establishment request from a client, the data connection establishment request carrying information on a plurality of receiving ports that have been enabled by the target device; and transferring data from the source device to the target device via the plurality of data connections, and data transfer/receiving method and apparatus according to the embodiments of the present disclosure leverage a plurality of parallel data connections from a source device to a target device to transfer data, thereby enhancing the speed and reliability of data backup or restoration.

Abstract: An online tree quota check tool computes quota usage of files in a directory tree while the structure of the directory tree keeps changing due to concurrent write access by other applications and clients. The quota check tool computes the quota usage without interfering with the ongoing file access operations that change the directory content and quota usage. A quota check database keeps track of which files have been checked and which have not, so that each file is checked and its quota usage is accumulated once and only once while directories are moved and directory scans are restarted as a result of the changing directory content. File system operations are modified to synchronize with the online tree quota check. The quota check tool can be single threaded or multi-threaded. A multi-threaded tool dispatches idle directory iteration threads to scan subdirectories in the directory tree.

Abstract: For automatically committing a file of a file system to a WORM committed state when the file has not been modified for a certain period of time, a file server maintains a list of file identifiers and periodically scans the list to check whether or not the certain period of time has expired for each file on the list. If so, the file is committed to the WORM state. The file server also enforces an in-place lock in response to a client request to modify or delete a file when the certain period of time is found to have expired for the file. A minimum modify time and a maximum modified time are included in each entry of the list of file identifiers in order to skip entries and to reclaim memory of entries during the periodic scan.