Abstract: An archiving file system is specifically designed to support the storage of, and access to, remote files stored on high speed, large capacity network data servers. The archiving file system automatically archives remote files across multiple types of secondary storage media on such network data servers based on a set of hierarchically selectable archival attributes selectively assigned to each remote file. The archiving file system is completely transparent to the user program and operates on remote files by providing a different file control program and a different file structure on the network data server, without the need to modify the standard file system that is native to a particular operating system program executing on the user nodes or the standard network file interfaces executing on the distributed computer network environment.

Abstract: A method and apparatus for backing up the control information of a file system to a secondary storage system provides for a fast and reliable recovery of the file system in the event of an unscheduled hard stop of a computer processing system. The file system utilizes control information that is maintained in a cache memory of the computer processing system and a copy of the control information is periodically backed up to a first and second logical device in the secondary storage system. As part of each backup, a control stamp value unique to each iteration of the backup is written to a pair of unique control stamp locations on the logical devices, one control stamp location being written prior to the back up of the control information and the other control stamp location being written after the backup of the control information.

Abstract: A method and apparatus for file storage allocation for secondary storage systems for allocating storage space for both small and large files in the same file system that dynamically utilizes at least two different sizes of logical storage allocation units to organize files stored on one or more physical storage devices. A first, smaller logical allocation unit is used to allocate space for the beginning of files until the size of the file passes a predefined maximum small allocation unit size. Beyond the predefined maximum small allocation unit size for a file, a second, larger logical allocation unit is used to allocate the remaining space necessary to store the file. In the preferred embodiment, small allocation units of 1K and large allocation units of 16K are used to map files directly and indirectly to the physical storage devices in a secondary file storage system.

Abstract: A pipelined data server having an improved data transfer architecture is used with a distributed computer network and a plurality of secondary storage devices to efficiently transfer data between the network and the secondary storage devices. The pipelined, multiprocessor data server includes a common inter-processor bus that connects one or more communication processors and file processors to one or more device processors, each having a buffer memory as part of the device processor. The common bus provides for global direct access to each of the buffer memories in the device processors by any of the other processors. The buffer memories are also connected to the secondary storage device attached to the device processor by a DMA transfer path in the device processor. In this way, data transfers can occur between the secondary storage device and the network with only one data transfer across the common bus.