-
公开(公告)号：KR1020080050980A
公开(公告)日：2008-06-10
申请号：KR1020070080134
申请日：2007-08-09
Applicant: 한국전자통신연구원
IPC: G06F15/16 , G06F15/173 , G06F17/00
Abstract: A method for providing a snapshot of a file set in an object based file system is provided to generate, eliminate or access the snapshot of the file set consistently while separating the snapshot information into a metadata server and an object storage server for providing the file set level snapshot in an object based file system which stores and manages the file as a logical object. A method for providing a snapshot of a file set comprises the following several steps. An object based file system locks all the files forming the file set before generating the snapshot(201). The object based file system guarantees consistency of the snapshot by locking the snapshot until finishing generating the snapshot(202). The system generates a snapshot file set(203). The system transmits a snapshot partition generation request to an object storage server(204). The system copies directory entries on all the files forming the file set to the snapshot file set(205). The system makes i-nodes of all the files of the file set point to a super block of the currently generated snapshot(206). The system maintains a snapshot file set list according to a snapshot performing order(207). If finishing generating the snapshot, the system gives back lock of the snapshot(208) and gives back lock of files(209).
Abstract translation: 提供了一种用于在基于对象的文件系统中提供文件集的快照的方法，用于在将快照信息分离成元数据服务器和用于提供文件集的对象存储服务器的同时，一致地生成，消除或访问文件集的快照 基于对象的文件系统中的高级快照，将文件作为逻辑对象存储和管理。 用于提供文件集的快照的方法包括以下几个步骤。 基于对象的文件系统在生成快照之前锁定形成文件集的所有文件（201）。 基于对象的文件系统通过锁定快照直到完成生成快照（202）来保证快照的一致性。 系统生成快照文件集（203）。 系统向对象存储服务器发送快照分区生成请求（204）。 系统将形成文件集的所有文件的目录条目复制到快照文件集（205）。 系统使文件集的所有文件的i节点指向当前生成的快照的超级块（206）。 系统根据快照执行顺序维护快照文件集列表（207）。 如果完成生成快照，系统会回放快照（208）的锁定并返回文件锁定（209）。
-
公开(公告)号：KR100825720B1
公开(公告)日：2008-04-29
申请号：KR1020060044257
申请日：2006-05-17
Applicant: 한국전자통신연구원
CPC classification number: G06F17/301 , G06F17/3012
Abstract: 파일 시스템에서 파일 관리 방법 및 이를 위한 메타데이터 서버가 개시된다. 본 발명의 파일 시스템에서 파일 관리 방법은 제1파일에 대해 동일한 제1식별자를 갖는 제1메타데이터 및 제1객체를 생성하는 단계; 및 생성된 복수의 파일들중 찾고자하는 이름의 제2파일에 대응되는 제2메타데이터 및 제2객체를 제2식별자를 이용하여 찾고, 제2메타데이터 및 제2객체가 찾아지면 제2파일을 개방하는 단계를 포함함을 특징으로 한다.
-
公开(公告)号：KR1020070061088A
公开(公告)日：2007-06-13
申请号：KR1020060044257
申请日：2006-05-17
Applicant: 한국전자통신연구원
CPC classification number: G06F17/301 , G06F17/3012
Abstract: A method for managing files in a file system and a metadata server for the same are provided to manage the files in a network-based asymmetric distributed file system using an OSD by checking and restoring reference consistency with the OSD following SCSI(Small Computer System Interface)/OSD protocol. A generator(121) generates metadata and an object of the file to be generated. A storing part(125) stores the metadata. A storage manager(122) stores the metadata to the storing part, stores the object to the outside, and manages the metadata and the object. A failure restorer(123) detects and restores en error generated in the metadata or the object when the generator generates the metadata/object or the storage manager manages the metadata/object. The generator, the storage manager, and the failure restorer communicate with the outside through an SCSI/OSD protocol to manage the object stored in the outside. The generator generates the metadata and the object having the same ID for the file.
Abstract translation: 提供用于管理文件系统中的文件和用于其的元数据服务器的方法，以通过使用OSD来管理基于网络的非对称分布式文件系统中的文件，所述方法通过在SCSI（小型计算机系统接口）之后检查并恢复与OSD的参考一致性 ）/ OSD协议。 生成器（121）生成要生成的文件的元数据和对象。 存储部分（125）存储元数据。 存储管理器（122）将元数据存储到存储部分，将对象存储到外部，并管理元数据和对象。 当生成器生成元数据/对象或存储管理器管理元数据/对象时，故障恢复器（123）检测并恢复在元数据或对象中生成的错误。 发生器，存储管理器和故障恢复器通过SCSI / OSD协议与外部通信，以管理存储在外部的对象。 生成器生成元数据，对象具有相同的文件ID。
-
公开(公告)号：KR1020030056540A
公开(公告)日：2003-07-04
申请号：KR1020010086768
申请日：2001-12-28
Applicant: 한국전자통신연구원
IPC: G06F17/40
CPC classification number: G06F11/1435 , G06F11/1471 , Y10S707/99931 , Y10S707/99953
Abstract: PURPOSE: A method for deleting and recovering a file preparing for a system failure on a DBMS(DataBase Management System) is provided to destroy the file safely and record a process preparing for the system failure generated during a file deleting process, and to recover a uncompleted file return error on the file return formed by repeated non-atomic operations during the system failure. CONSTITUTION: If the file to be deleted is found in a pending action list, the atomic recording of each allocation unit for the non-atomic file deletion is performed(S802). An address of the last log record at the time is stored in a main memory by setting up the NTA after receiving the allocation of the file to be deleted from the list(S802). The atomic recording steps(S811-814) of the allocation unit are repeated until no more return allocation unit is found in the deleted file. If all allocation units are returned, a file deletion log including a file discriminator is recorded to the log, and the corresponding discriminator is returned to the system(S815).
Abstract translation: 目的：提供一种在DBMS（DataBase管理系统）上删除和恢复准备系统故障的文件的方法，以安全地销毁该文件，并记录准备在文件删除过程中生成的系统故障的进程，并恢复 在系统故障期间通过重复的非原子操作形成的文件返回形式的未完成文件返回错误。 构成：如果待处理的动作列表中找到要删除的文件，则执行非原子文件删除的每个分配单元的原子记录（S802）。 通过在从列表接收到要删除的文件的分配之后设置NTA，将此时的最后日志记录的地址存储在主存储器中（S802）。 重复分配单元的原子记录步骤（S811-814），直到在删除的文件中没有找到更多的返回分配单元。 如果返回所有分配单元，则将包括文件鉴别器的文件删除日志记录到日志中，并将对应的鉴别器返回给系统（S815）。
-
-
Patent Agency Ranking
-
-
公开(公告)号：KR101733118B1
公开(公告)日：2017-05-08
申请号：KR1020110084634
申请日：2011-08-24
Applicant: 한국전자통신연구원
Abstract: 메타데이터연산충돌회피방법및 이를수행하는메타데이터관리시스템이개시된다. 메타데이터연산충돌회피방법은디렉토리메타데이터저장장치가미리저장된디렉토리메타데이터정보를참조하여요청된메타데이터연산의실행가능여부를판단하는단계와, 디렉토리메타데이터저장장치가상기요청된메타데이터연산이실행가능한경우요청된메타데이터연산에상응하는메타데이터연산의도를특정파일메타데이터저장장치에제공하는단계와, 특정파일메타데이터저장장치가미리저장된파일메타데이터정보또는미리저장된메타데이터연산의도정보를참조하여메타데이터연산의도에포함된요청된메타데이터연산의실행가능여부를판단하는단계와, 특정파일메타데이터저장장치가요청된메타데이터연산이실행가능한경우메타데이터연산의도를기록하고, 메타데이터연산의도기록완료정보를디렉토리메타데이터저장장치에제공하는단계및 디렉토리메타데이터저장장치가특정파일메타데이터저장장치로부터제공된메타데이터연산의도기록완료정보에상응하여요청된메타데이터연산을실행하는단계를포함한다. 따라서, 파일시스템의일관성을손상시킬수 있는메타데이터연산을방지할수 있다.
-
公开(公告)号：KR101713314B1
公开(公告)日：2017-03-07
申请号：KR1020130049990
申请日：2013-05-03
Applicant: 한국전자통신연구원
CPC classification number: G06F17/30117
Abstract: 분산네트워크시스템에서가비지파일을완전하게삭제할수 있는방법및 시스템이제공된다. 본발명에따르면데이터서버에접근할수 없어삭제하려는데이터를삭제하지못하여찌꺼기파일이생성된경우, 생성된찌꺼기파일까지완전하게삭제할수 있다. 이때, 찌꺼기파일에대한삭제연산을분산된데이터서버단위로수행함으로써연산효율성을극대화할수 있다.
Abstract translation: 提供了可以完全删除分布式网络系统中的垃圾数据的方法和系统。 由于不可能初始访问数据服务器，所以要删除的数据不会被删除，因此当生成垃圾文件时，可以完全删除生成的垃圾文件。 在这种情况下，通过在分散数据服务器单元中执行垃圾文件的删除操作，可以使操作效率最大化。
-
公开(公告)号：KR101694984B1
公开(公告)日：2017-01-11
申请号：KR1020100125106
申请日：2010-12-08
Applicant: 한국전자통신연구원
CPC classification number: G06F11/1076 , G06F2211/1028 , G06F2211/1059
Abstract: 본발명은비대칭클러스터링파일시스템에서의패리티산출방법에관한것으로, 클라이언트시스템에서저장하고자하는파일을기 설정된크기의청크로분할하여복수의데이터서버로분산저장하는단계; 상기클라이언트시스템이메타데이터서버로데이터쓰기완료알림메시지를전송하는단계; 상기메타데이터서버에서상기데이터쓰기완료알림메시지를통해전달된파일을복구큐에저장하는단계; 및상기메타데이터서버가상기복구큐에저장된파일에대한패리티산출요청을임의의데이터서버로전송하는단계를포함하여구성된다.
Abstract translation: 公开了一种计算奇偶校验的方法。 该方法由客户端系统划分要被存储为预定大小的块并将多个数据分组存储在多个数据服务器中的文件; 由客户端系统向元数据服务器发送数据写入完成通知消息; 由元数据服务器将通过数据写入完成通知消息发送的文件存储在恢复队列中; 并且由元数据服务器向随机数据服务器发送存储在恢复队列中的文件的奇偶校验计算请求。
-
-
-
-
-
-
-
-
-
Search Results
98
records
(took 0.023 seconds)
