Abstract:
PURPOSE: A method for first outputting shared data using a global lock pattern on a shared file system is provided to reduce a process delay time of a file system by previously reading and offering the shared data from a memory of other host or a disk of the shared file system when the global lock for reading and writing the shared data is requested from the host. CONSTITUTION: The shared file system of the host requests the global lock for the shared data through a global lock manager. The global lock manager finds out a block allotting the shared data requested from the host by receiving the global lock request. The global lock manager increases a count for the allotted block of the shared data requested by the host one by one. The global lock manager reads the shared data from the memory or the disk of the host having the allotted block of the shared data. The shared data is transmitted to the host.
Abstract:
PURPOSE: A method for journaling and recovering a shared disk file system is provided to increase a journaling speed by transmitting the changed metadata through network without reflecting to a disk if one host changes the metadata and other host accesses the same metadata in a shared disk environment such as an SAN(Storage Area Network) environment. CONSTITUTION: If a system call operation for processing journaling is occurred, a new transaction region is allotted and initialized by staring a transaction for guaranteeing the recovery, and a transaction type is set. The transaction manages the lock information by getting the lock information for the changed information and adding it to the transaction, and the update of the changed metadata is pinned while the metadata is changed(201). The changed metadata added to the transaction and the changed information for the important general data are stored in a journal space, and the lock information connected to the transaction is unlocked(203).
Abstract:
PURPOSE: A RAID(Redundant Arrays of Inexpensive Disks) sub system and a method for inputting/outputting and recovering data on a degraded mode by using the same are provided to increase a response speed of a system on the degraded mode, and to decrease the system load according to operation by inputting/outputting the data requested from a host computer as using a block of a sparing disk drive. CONSTITUTION: The system comprises a disk array controller(200) connecting to a host terminal(100) and a disk array(210) through a bus, and the disk array(210) including disk drives(211-214) and the sparing disk drive(215). The disk array controller(200) distributes and stores the data in the disk drives(211-214), processes the input/output parallel, and processes the input/output of the data to be inputted/output from the degraded disk through the sparing disk drive(215). The sparing disk drive(215) includes a sparing disk recovery bitmap checking the recovery of the sparing disk blocks, and recovers the block of the degraded disk drive according to the data input/output request of the host computer(100).
Abstract:
PURPOSE: A method for managing a logical volume for minimizing the dynamic size change and a metadata volume is provided to manage the logical volume by using a mapping table for a relation between a logical and a physical address and a minimal storage for the system metadata. CONSTITUTION: The logical volume is generated by gathering the disk partitions according to a logical volume configuration request for a physical storage. The metadata including the information for the disk partitions joining with the logical and the physical volume and stores the metadata in the disk partition. A size of the logical volume is dynamically changed by receiving a change request from the disk partition configuring the logical volume and the change of the metadata needed to change the logical volume is carried out. The physical address corresponding to the mapping information of the metadata storing the information of the physical address matched with the logical address is understood and returned by receiving the logical address of the logical volume.
Abstract:
PURPOSE: A large-capacity file storage system and a method of adding and deleting data blocks of dynamic multi-level inode are provided to effectively manage large-capacity files. CONSTITUTION: A large-capacity file storage system includes a pointer having a level different from the level of an inode information area(101A) storing information about root inode. The pointer includes a double indirect inner pointer in which a data block exists through double indirect pointer nodes, a single indirect inner pointer in which a data block exists through a single indirect pointer node, and a direct point that directly points a data block. The pointer of the root inode is allocated according to the direct pointer, single indirect pointer node and double indirect pointer node by increasing and decreasing the level in accordance with the size of the data block.
Abstract:
PURPOSE: A concurrency controlling method is provided to enhance a concurrency of a search or an insertion operation by updating a split and an MBR(Master Boot Record) operation so that it can satisfy a request condition of a commercial DBMS(Database Management System), for example, a contents based image search. CONSTITUTION: The method comprises steps of finding a terminal node appropriate for inserting a new entry into(S501), giving back a path stack, storing the path from a root to a found terminal node, an exclusive lock and a shared latch on the found terminal node, and checking if there exists a space at the found node for inserting the new entry into(S502), releasing a shared latch of the terminal node and obtaining an exclusive latch(S503), inserting the entry into the found terminal node(S504), checking if an MBR is changed by the insertion of the entry(S505), releasing the exclusive latch of the terminal node(S506), reflecting the release of the exclusive latch of the terminal node on ancestor nodes stored at the path stack(S507), releasing all the latches and locks(S508), in a case that there is no space in the step S502, obtaining a tree lock, recording an NTA start, performing a split operation, recording a dummy CLR and releasing all the latches and locks(S509, S510, S511, S512, S508).
Abstract:
현재, 다중 에이전트 관리기 개발은 크게 두 방향 비표준화 방향과 표준화 방향으로 진행되고 있다. 비표준화 방향은 다른 다중 에이전트 시스템과의 호환성을 고려하지 않고 단지 개발중인 시스템 자체에 초점을 맞추어 독자적인 모델로 다중 에이전트 관리기를 개발하는 것이고, 표준화 방향은 다른 에이전트 시스템과의 호환성을 고려하여 현재 진행중인 표준화 권고안에 따라 다중 에이전트 관리기를 개발하는 것이다. 비표준화 개발의 문제점은 호환성이 결여되어있다는 것과 특정 시스템에 의존적으로 개발되고 있으므로 다른 시스템 개발 시 반복적으로 재개발되어야 한다는 것이다. 표준화 개발의 문제점은 물리적으로 모든 에이전트들이 구동 되어 있다는 것을 전제로 하고 논리적으로는 이들간의 다양한 구성으로 에이전트 기반 서비스를 지원하므로 자원 관리의 효율성에 있어서 문제점을 가지고 있다. 본 발명에서는 표준화를 고려하면서 동시에 자원 관리의 효율성을 증대시키는 다중 에이전트 동적 관리기에 관하여 기술한다. 다중 에이전트 동적 관리기는 에이전트 표준화 단체인 FIPA(Foundation Intelligent Physical Agent)의 권고안에 따라 구성하여 타 에이전트 시스템과의 호환성을 해결하였으며, 필요한 시점에 필요한 에이전트들을 구동시키는 동적인 에이전트 프로세스 제어 방법을 채택함으로써 자원 관리의 효율성을 가져왔다.
Abstract:
본 발명은 컴퓨터를 사용한 데이터 회의 시스템에서 회의에 참여하고 있는 응용 프로그램들간의 능력을 실시간으로 협상해 주는 응용 프로그램 능력 협상 방법에 관한 것이다. 기존의 데이터 회의 시스템에서 응용 프로그램들이 정해진 능력으로만 서로 통신할 수 있게 되어 있어 특정 성능 이상의 시스템에서만 통신을 할 수 있었다. 또한 회의에 참여한 응용 프로그램들도 다른 응용 프로그램의 능력을 알 수 없어 정해진 능력보다 훨씬 나은 성능으로 통신할 수 있음에도 불구하고 정해진 능력으로 통신할 수 밖에 없었다. 본 발명은 이러한 문제점을 해결하기 위해서 진행중인 회의에서 어떤 응용 프로그램이 참여하거나 탈퇴 했을 경우 이 응용 프로그램과 통신하는 다른 동료 응용 프로그램들과 능력 협상을 하여 이들간에 최상의 조건으로 통신할 수 있는 능력 값을 계산한 후 이를 회의에 참여한 모든 노드에게 알려 줌으로써 언제든지 최상의 상태로 통신할 수 있게 하여 준다.