[go: up one dir, main page]

CN1255748C - Metadata hierarchy management method and system of storage virtualization system - Google Patents

Metadata hierarchy management method and system of storage virtualization system Download PDF

Info

Publication number
CN1255748C
CN1255748C CN 200310111436 CN200310111436A CN1255748C CN 1255748 C CN1255748 C CN 1255748C CN 200310111436 CN200310111436 CN 200310111436 CN 200310111436 A CN200310111436 A CN 200310111436A CN 1255748 C CN1255748 C CN 1255748C
Authority
CN
China
Prior art keywords
metadata
server
list
directory
matching
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN 200310111436
Other languages
Chinese (zh)
Other versions
CN1545047A (en
Inventor
金海�
贾永洁
吴松
冉龙波
王志平
周润松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN 200310111436 priority Critical patent/CN1255748C/en
Publication of CN1545047A publication Critical patent/CN1545047A/en
Application granted granted Critical
Publication of CN1255748C publication Critical patent/CN1255748C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明公开了一种存储虚拟化系统的元数据层次管理方法及其系统。该方法对元数据逻辑树进行层次管理,引入匹配表以减轻根目录服务器的压力,不仅可以快速准确的定向到用户所请求访问的目录,而且使系统达到更好的扩展性。匹配表常驻内存,因为它只记录每个目录服务器的根入口地址,所以所占空间极小,而且管理方便。由于匹配表所拥有的条目并不多,搜索匹配表对用户请求目录进行匹配的效率也很高。其系统包括设置有匹配表及匹配表管理模块的元数据服务器和目录服务器;匹配表至少包括目录服务器名称和目录服务器上所保存的元数据子逻辑树的根目录。本发明具有效率高、扩展性好和易于管理等特点。

Figure 200310111436

The invention discloses a metadata layer management method of a storage virtualization system and a system thereof. The method manages the logical tree of metadata hierarchically, introduces a matching table to reduce the pressure on the root directory server, not only can quickly and accurately direct to the directory requested by the user, but also enables the system to achieve better scalability. The matching table is resident in memory, because it only records the root entry address of each directory server, so it occupies a very small space and is easy to manage. Since the match table does not have many entries, searching the match table to match the user request directory is also very efficient. The system includes a metadata server and a directory server provided with a matching table and a matching table management module; the matching table at least includes the name of the directory server and the root directory of the metadata sub-logic tree stored on the directory server. The invention has the characteristics of high efficiency, good expansibility, easy management and the like.

Figure 200310111436

Description

一种存储虚拟化系统的元数据层次管理方法及其系统Metadata level management method and system for storage virtualization system

技术领域technical field

本发明属于计算机存储领域,具体涉及一种存储虚拟化系统的元数据层次管理方法及其系统。The invention belongs to the field of computer storage, and in particular relates to a metadata layer management method of a storage virtualization system and a system thereof.

背景技术Background technique

在现代高性能计算中,科学计算和军事应用对存储的要求越来越高,包括大容量、分布式、高性能和高可靠性。广域网的虚拟化存储管理方式将地理上分布的各种高性能存储系统集成为一体,形成庞大的分布存储空间,充分实现资源共享,提高资源利用率,有效解决存储数据的爆炸性增长和存储管理能力相对不足之间的矛盾。In modern high-performance computing, scientific computing and military applications have increasingly high requirements for storage, including large capacity, distributed, high performance, and high reliability. The virtual storage management method of WAN integrates various high-performance storage systems that are geographically distributed to form a huge distributed storage space, fully realize resource sharing, improve resource utilization, and effectively solve the explosive growth of stored data and storage management capabilities The contradiction between relative inadequacy.

在一个广域存储系统中,这些海量的存储资源必须被有效的管理,从而引入元数据的概念。元数据是描述数据的数据,它为系统提供对象物理位置与其逻辑名字之间的映射,一个逻辑文件可以对应多个物理文件副本。此外,元数据还包括文件目录信息,文件信息,存储设备信息及相关的系统信息等等。In a wide-area storage system, these massive storage resources must be managed effectively, thus introducing the concept of metadata. Metadata is the data that describes the data. It provides the system with a mapping between the physical location of the object and its logical name. A logical file can correspond to multiple physical file copies. In addition, metadata also includes file directory information, file information, storage device information and related system information, etc.

在许多系统中比如SRB或者GridFTP,都采用层次目录结构管理元数据。当元数据变得庞大时,系统会同时启用多个元数据服务器,这样会导致以下问题的产生:一方面,逻辑树必须分布在这些元数据服务器上,而另一方面,目录服务器之间必须协同操作,因为要返回的元数据可能分布在不同的元数据服务器上,这样就对元数据的根目录服务器造成了压力。第三,如果根目录服务器出现故障,整个目录服务系统也就不能再正常运转了,很难做到系统的高可用。最后,这种结构使得元数据服务器的扩展也变得很困难。In many systems such as SRB or GridFTP, a hierarchical directory structure is used to manage metadata. When the metadata becomes huge, the system will enable multiple metadata servers at the same time, which will lead to the following problems: on the one hand, the logical tree must be distributed on these metadata servers, and on the other hand, the directory servers must be Collaborative operation, because the metadata to be returned may be distributed on different metadata servers, which puts pressure on the root directory server of metadata. Third, if the root directory server fails, the entire directory service system will no longer be able to function normally, and it is difficult to achieve high availability of the system. Finally, this structure makes it difficult to scale the metadata server.

发明内容Contents of the invention

本发明的目的在于提供一种能克服上述缺陷的存储虚拟化系统的元数据层次管理方法。该方法具有更高的效率、更好的扩展性,并且更易于管理。本发明还提供了实现该方法的系统。The object of the present invention is to provide a metadata level management method of a storage virtualization system that can overcome the above defects. This approach is more efficient, scales better, and is easier to manage. The invention also provides a system for realizing the method.

本发明提供的一种存储虚拟化系统的元数据层次管理方法,其步骤为:A metadata level management method for a storage virtualization system provided by the present invention, the steps of which are as follows:

(1)把整个存储虚拟化系统划分为多个逻辑域,每个逻辑域内的元数据分布在多个目录服务器上;(1) Divide the entire storage virtualization system into multiple logical domains, and the metadata in each logical domain is distributed on multiple directory servers;

(2)在各逻辑域中设置一个元数据服务器,用于管理该域中的目录服务器;(2) Set up a metadata server in each logical domain to manage the directory server in the domain;

(3)在各元数据服务器中设置一个匹配表,该匹配表至少包括二个字段:目录服务器名称和目录服务器上所保存的元数据子逻辑树的根目录,用于直接将用户请求定向到其所在的目录服务器;(3) A matching table is set in each metadata server, and the matching table includes at least two fields: the root directory of the metadata sub-logic tree stored on the directory server name and the directory server, for directly directing user requests to the directory server on which it resides;

(4)当用户提出请求时,从用户请求中提取要访问目录的路径;(4) When the user makes a request, extract the path of the directory to be accessed from the user request;

(5)在该域的匹配表中对路径进行长匹配,判断在匹配表中是否存在符合的目录服务器;(5) carry out long match to path in the matching table of this domain, judge whether there is the directory server that conforms to in the matching table;

(6)如果存在符合的目录服务器,直接访问匹配的目录服务器,并返回结果;否则显示出错信息。(6) If there is a matching directory server, directly access the matching directory server and return the result; otherwise, an error message will be displayed.

实现上述方法的系统,被划分为多个逻辑域,每个逻辑域内的元数据分布在多个目录服务器上,其特征在于:每个逻辑域指定一台服务器作为元数据服务器,指定多台服务器作为保存元数据信息的目录服务器;元数据服务器内设置有匹配表及匹配表管理模块,匹配表至少包括目录服务器名称和目录服务器上所保存的元数据子逻辑树的根目录,匹配表管理模块用于匹配表中的记录进行维护、匹配表文件的解析以及根据用户请求的路径匹配到合适的目录服务器或显示出错信息。The system for implementing the above method is divided into multiple logical domains, and the metadata in each logical domain is distributed on multiple directory servers. It is characterized in that: each logical domain specifies one server as the metadata server, and multiple servers As a directory server for storing metadata information; a matching table and a matching table management module are arranged in the metadata server, the matching table includes at least the name of the directory server and the root directory of the metadata sub-logic tree stored on the directory server, and the matching table management module It is used for maintaining the records in the matching table, parsing the matching table file, and matching to the appropriate directory server or displaying error messages according to the path requested by the user.

本发明的元数据层次管理系统具有以下优点及效果。The metadata hierarchical management system of the present invention has the following advantages and effects.

(1)更高的效率:匹配表是常驻内存的,它保存的条目数受域内目录服务器数目所限,所以遍历匹配表对用户请求目录进行匹配的速度也很快。遍历后可以直接访问请求目录所在的目录服务器,而不用再途经根目录目录服务器,这将大大减轻根目录所在的目录服务器的负载,提高元数据的搜索的速度继而提高整个系统的效率。(1) Higher efficiency: the matching table is resident in memory, and the number of entries it saves is limited by the number of directory servers in the domain, so the speed of traversing the matching table to match user requested directories is also very fast. After traversing, you can directly access the directory server where the requested directory is located, without going through the root directory server, which will greatly reduce the load on the directory server where the root directory is located, improve the speed of metadata search and then improve the efficiency of the entire system.

(2)更好的扩展性:分层的树型管理模式使维护和管理易于扩展;当系统负载上升需要增加目录服务器时,只需要在匹配表中增加几个条目,标示出系统中新增加的目录服务器及其对应的根目录即可。因为有了匹配表的调度,目录服务器可以达到相互独立,不需要感知域内其他目录服务器的存在。这样,系统效率不会随目录服务器数目的增多而下降,达到系统效率的可扩展性。(2) Better scalability: The hierarchical tree management mode makes maintenance and management easy to expand; when the system load increases and the directory server needs to be added, only a few entries need to be added to the matching table to mark the new additions in the system directory server and its corresponding root directory. Because of the scheduling of the matching table, the directory servers can be independent from each other, and do not need to be aware of the existence of other directory servers in the domain. In this way, the system efficiency will not decrease with the increase of the number of directory servers, and the scalability of the system efficiency is achieved.

(3)易于管理:匹配表的结构很简单,只记录了每个目录服务器的名字及存放在其上的元数据的根目录。所以其添加删除修改都很方便。(3) Ease of management: the structure of the matching table is very simple, only recording the name of each directory server and the root directory of the metadata stored thereon. So it is very convenient to add, delete and modify.

当发明采用JAVA作为开发工具时,可以实现平台无关性;当存储虚拟化系统增加元数据副本时,由于副本分布情况能够完全在匹配表中显示出来,可以通过匹配表选择一个合适的目录服务器,从而有效的利用副本,提高系统效率。When the invention uses JAVA as a development tool, platform independence can be achieved; when the storage virtualization system adds metadata copies, since the copy distribution can be completely displayed in the matching table, a suitable directory server can be selected through the matching table, In this way, copies can be effectively utilized and system efficiency can be improved.

附图说明Description of drawings

图1为域内元数据逻辑树示例图;Figure 1 is an example diagram of a logical tree of metadata in a domain;

图2为本发明的元数据访问流程;Fig. 2 is the metadata access process of the present invention;

图3为现有元数据逻辑树的访问方式;Fig. 3 is the access method of the existing metadata logic tree;

图4为本发明元数据逻辑树的访问方式;Fig. 4 is the access mode of the logical tree of metadata of the present invention;

图5为元数据管理的结构图。Fig. 5 is a structural diagram of metadata management.

具体实施方式Detailed ways

在我们的存储虚拟化系统GSP(G1obal Storage Provider)中,通过划分逻辑域对整个系统进行层次管理。在每个域内,元数据分布在多个目录服务器上而,并按照类似Linux文件系统的组织方式形成一棵“本域元数据逻辑树”,图1是一个域元数据逻辑树示例。图中,1.1,1.2,1.3,1.4表示域内的目录服务器;A0表示域内元数据逻辑树的根目录;A1,A2…,表示域内元数据逻辑树的子目录。In our storage virtualization system GSP (Global Storage Provider), the entire system is managed hierarchically by dividing logical domains. In each domain, metadata is distributed on multiple directory servers, and forms a "local domain metadata logical tree" in an organization similar to the Linux file system. Figure 1 is an example of a domain metadata logical tree. In the figure, 1.1, 1.2, 1.3, 1.4 represent the directory servers in the domain; A0 represents the root directory of the metadata logical tree in the domain; A1, A2..., represent the subdirectories of the metadata logical tree in the domain.

每个用户都隶属于一个域,用户的根目录绑定到域内一个子目录上,用户登陆时,系统自动定向到用户的根目录去。Each user belongs to a domain, and the user's root directory is bound to a subdirectory in the domain. When the user logs in, the system automatically directs to the user's root directory.

建立基于目录树结构的元数据目录服务器(Directory Server)来存储元数据及其副本。但是在广域网的范围内,随着系统中资源的增长,文件和目录的信息会变得越来越庞大,很显然,目录服务器组织得不好,就会成为系统的瓶颈。所以如何快速准确的定向到用户的根目录就是一个十分突出的问题,这将影响到元数据的搜索速度进而影响整个系统的效率。在我们的系统中,引入匹配表来解决上述问题。Establish a metadata directory server (Directory Server) based on the directory tree structure to store metadata and its copies. However, within the scope of the wide area network, with the increase of resources in the system, the information of files and directories will become larger and larger. Obviously, if the directory server is not well organized, it will become the bottleneck of the system. So how to quickly and accurately direct to the user's root directory is a very prominent problem, which will affect the search speed of metadata and then affect the efficiency of the entire system. In our system, matching tables are introduced to solve the above problems.

在每个域中设置一个元数据服务器管理域中的多个目录服务器,匹配表保存在元数据服务器上。匹配表是一个常驻内存的数据结构,记录了每个目录服务器的名字及存放在其上的元数据的根目录;如果一个目录服务器上面保存了某个目录的副本,也把这个目录放到匹配表中去。在匹配表中,还可以设置索引字段,作为某个条目在匹配表中的唯一标识。A metadata server is set up in each domain to manage multiple directory servers in the domain, and the matching table is saved on the metadata server. The matching table is a data structure resident in memory, which records the name of each directory server and the root directory of the metadata stored on it; if a copy of a directory is saved on a directory server, this directory is also placed in the Match table to go. In the match table, an index field can also be set as a unique identifier of an entry in the match table.

当系统收到用户的请求时,首先在元数据服务器上遍历匹配表,根据用户所请求访问的路径对匹配表包含的所有路径做一个最长匹配,找到用户要访问的目录所在的目录服务器,直接访问这个目录服务器,而不用通过域内元数据根目录所在的目录服务器依次查找;如果元数据存在副本,还可以从匹配表中选择一个合适的目录服务器给用户访问。元数据访问流程如图2所不。When the system receives a request from a user, it first traverses the matching table on the metadata server, performs a longest match on all paths contained in the matching table according to the path requested by the user, and finds the directory server where the directory the user wants to access is located. Directly access this directory server instead of searching through the directory servers where the metadata root directory in the domain is located; if there is a copy of the metadata, you can also select a suitable directory server from the matching table for the user to access. The metadata access process is shown in Figure 2.

图1列举了一个逻辑树的例子。这个逻辑树代表一个域内的文件目录结构,分布在四个目录服务器上。其中,逻辑树的根/A0存放在目录服务器1.1上,目录/A0/B1存放在目录服务器1.2上,目录/A0/A1/A2/C11和目录/A0/A1/A2/C12放在目录服务器1.3上,目录/A0/B1/B2/B3/B4/D1放在目录服务器1.4上。表2显示了与图1逻辑树对应的匹配表。Figure 1 shows an example of a logic tree. This logical tree represents the file directory structure within a domain and is distributed across four directory servers. Among them, the root /A0 of the logical tree is stored on the directory server 1.1, the directory /A0/B1 is stored on the directory server 1.2, the directory /A0/A1/A2/C11 and the directory /A0/A1/A2/C12 are stored on the directory server On 1.3, the directory /A0/B1/B2/B3/B4/D1 is placed on directory server 1.4. Table 2 shows the matching table corresponding to the logic tree in Fig. 1.

假设用户想访问/A0/B1/B2/B3/B4/D1/D2,如果按照现存的元数据访问方式,首先会从域内逻辑树的根目录所在的目录服务器1.1开始查找,历经目录服务器1.3,到达目录服务器1.4,然后再把结果返回给用户,如图3所示。图中,R表示用户请求访问的文件/目录的路径;1.1,1.2,1.3,1.4表示本域中的目录服务器;A0表示域内元数据逻辑树的根目录;B1,D1…,表示域内元数据逻辑树中的目录;虚线箭头和r1,r2…表示用户访问步骤。但是如果采取本发明所用的匹配表方式,系统收到用户的请求后,将首先查找匹配表,根据用户请求的路径做一个最长匹配,这样就可以得到元数据在目录服务器1.4上面。系统将直接访问目录服务器1.4,而不需要再通过根目录入口1.1,如图4所示。图中,1表示域内的所有的目录服务器,2表示域内元数据服务器GNS,3表示匹配表,虚线箭头和r1,r2表示用户访问步骤。Assuming that the user wants to access /A0/B1/B2/B3/B4/D1/D2, according to the existing metadata access method, it will first search from the directory server 1.1 where the root directory of the logical tree in the domain is located, and go through the directory server 1.3. Reach the directory server 1.4, and then return the result to the user, as shown in Figure 3. In the figure, R represents the path of the file/directory that the user requests to access; 1.1, 1.2, 1.3, 1.4 represent the directory server in the domain; A0 represents the root directory of the logical tree of metadata in the domain; B1, D1..., represent the metadata in the domain Directories in the logical tree; dotted arrows and r1, r2... denote user access steps. However, if the matching table mode used in the present invention is adopted, after the system receives the user's request, it will first search the matching table, and do a longest match according to the path requested by the user, so that metadata can be obtained on the directory server 1.4. The system will directly access the directory server 1.4 without going through the root directory entry 1.1, as shown in Figure 4. In the figure, 1 indicates all directory servers in the domain, 2 indicates the metadata server GNS in the domain, 3 indicates the matching table, dashed arrows and r1, r2 indicate user access steps.

为了达到有效搜索的目的,所有目录服务器都必须自己保存域内元数据树的逻辑结构。例如,目录服务器1.4仍然会保存域内元数据树的根,但它只是一个路径,并不保存根的内容,这样当一个搜索请求到达1.4的时候就不需要对请求做任何变换。这个根条目由于内容为空,只需要占用极少的存储空间,并且维护起来也很方便。In order to achieve the purpose of effective search, all directory servers must maintain the logical structure of the metadata tree in the domain. For example, directory server 1.4 still saves the root of the metadata tree in the domain, but it is just a path and does not save the content of the root, so that when a search request reaches 1.4, it does not need to make any changes to the request. Because the content of this root entry is empty, it only needs to occupy very little storage space, and it is also very convenient to maintain.

匹配表的内容存放在一个文件中,当整个系统启动的时候,把文件中保存的匹配表读入内存。匹配表的结构和功能决定了一般情况下都是在内存中对匹配表进行读操作,这就不会涉及到对匹配表文件的读写。只有在新增目录服务器或者现有目录服务器的根目录发生变化的情况下才对匹配表进行修改,进而对文件进行修改,并不需要频繁读写匹配表的文件,大大节省了访问匹配表的时间。The content of the matching table is stored in a file, and when the whole system is started, the matching table saved in the file is read into the memory. The structure and function of the matching table determine that the reading operation of the matching table is generally performed in the memory, which does not involve the reading and writing of the matching table file. Only when a new directory server is added or the root directory of an existing directory server changes, the matching table is modified, and then the file is modified. It does not need to read and write the files of the matching table frequently, which greatly saves the time of accessing the matching table. time.

在元数据服务器中添加一个匹配表管理模块,对匹配表中的记录进行维护,具体操作包括记录的添加、删除、更新操作,匹配表文件的解析以及根据用户请求的路径匹配到合适的目录服务器等等,具体解释如下。Add a matching table management module to the metadata server to maintain the records in the matching table. The specific operations include adding, deleting, and updating records, parsing the matching table file, and matching to the appropriate directory server according to the path requested by the user. Wait, the specific explanation is as follows.

匹配表记录的添加:当域内新增一个目录服务器的时候,向匹配表中添加一条记录;如果某个目录服务器中的一个子目录过大,需要迁移到另外一个目录服务器去,也要通知匹配表,增加一条相关记录,以记录这个子目录和其迁移到的目录服务器。Adding matching table records: When a directory server is added in the domain, a record is added to the matching table; if a subdirectory in a certain directory server is too large and needs to be migrated to another directory server, the matching table should also be notified Table, add a related record to record this subdirectory and the directory server it migrates to.

匹配表记录的删除:当某个目录服务器不再服务于某个域时,从匹配表上删除相应的记录;Deletion of matching table records: When a directory server no longer serves a certain domain, delete the corresponding record from the matching table;

匹配表记录的修改:当某个目录服务器上保存的目录树根目录发生变化时,修改相应的记录;Modification of matching table records: When the root directory of the directory tree stored on a directory server changes, modify the corresponding record;

匹配表文件的解析:当系统启动的时候,把匹配表文件中的内容读到字符串数组中,常驻内存;Analysis of the matching table file: when the system starts, read the contents of the matching table file into the string array, which is resident in memory;

匹配用户请求的路径:在匹配表中查找用户请求访问的路径所在的目录服务器。Match the path requested by the user: Find the directory server where the path requested by the user is located in the matching table.

使用集群系统中的5个节点构建一个元数据层次管理系统,其基本配置如表1所示。Use five nodes in the cluster system to construct a metadata hierarchical management system, and its basic configuration is shown in Table 1.

  CPU CPU   内存 Memory   硬盘 hard drive   网卡 network card   操作系统 operating system   网络 network   双PIII 866 Double PIII 866   256M 256M   30G 30G   3C905B 3C905B   Linux 6.2 Linux 6.2   100M交换机 100M switch

                  表1  各节点的硬件及网络配置Table 1 Hardware and network configuration of each node

其中,一台作为元数据服务器,另外4个作为目录服务器。元数据服务器负责匹配表的维护,包括生成添加删除更新等一系列的操作。目录服务器负责保存元数据信息。Among them, one acts as a metadata server, and the other four act as directory servers. The metadata server is responsible for the maintenance of the matching table, including a series of operations such as generating, adding, deleting, and updating. The directory server is responsible for storing metadata information.

具体实施如下:其中一个节点充当元数据服务器,装载保存匹配表的文件和匹配表管理模块;其余四个节点充当1.1,1.2,1.3,1.4,装载LDAP数据库,保存域内的元数据。The specific implementation is as follows: one of the nodes acts as a metadata server, and loads the file for saving the matching table and the matching table management module; the other four nodes act as 1.1, 1.2, 1.3, and 1.4, loading the LDAP database, and storing metadata in the domain.

依照图5,4.1表示一个逻辑域Domain,其后的阴影部分4.2和4.3代表和4.1结构相同的多个逻辑域;5表示匹配表管理模块,虚线箭头表示元数据服务器和目录服务器之间的互访。According to Figure 5, 4.1 represents a logical domain Domain, and the shaded parts 4.2 and 4.3 represent multiple logical domains with the same structure as 4.1; 5 represents the matching table management module, and the dotted arrow represents the interaction between the metadata server and the directory server visit.

整个系统的配置说明如下:The configuration of the whole system is described as follows:

(1)匹配表包括两个字段,其示例如表2。   根目录路径   目录服务器   /A0   1.1   /A0/B1   1.2   /A0/A1/A2/C11   1.3   /A0/A1/A2/C12   1.3   /A0/B1/B2/B3/B4/D1   1.4 (1) The matching table includes two fields, an example of which is shown in Table 2. root directory path directory server /A0 1.1 /A0/B1 1.2 /A0/A1/A2/C11 1.3 /A0/A1/A2/C12 1.3 /A0/B1/B2/B3/B4/D1 1.4

            表2  匹配表示例Table 2 Example of matching table

各字段解释如下:The fields are explained as follows:

路径:目录服务器保存的元数据子逻辑树的根目录路径;Path: the root directory path of the metadata sub-logical tree saved by the directory server;

目录服务器名称:此目录服务器的名字,域内各目录服务器名称唯一。Directory server name: the name of the directory server, each directory server name in the domain is unique.

(2)域内元数据保存在四个目录服务器上,具体分布示例如表3。   目录服务器名称   目录服务器保存的元数据信息   1.1   /A0下的元数据,但不包括目录/A0/B1,/A0/A1/A2/C11,/A0/A1/A2/C12和/A0/B1/B2/B3/B4/D1下的内容   1.2   /A0/B1下的元数据,但不包括目录/A0/B1/B2/B3/B4/D1下的内容   1.3   目录/A0/A1/A2/C11和/A0/A1/A2/C12下的元数据   1.4   目录/A0/B1/B2/B3/B4/D1下的元数据 (2) Metadata in the domain are stored on four directory servers, and the specific distribution example is shown in Table 3. directory server name Metadata information stored by the directory server 1.1 Metadata under /A0, but not under directories /A0/B1, /A0/A1/A2/C11, /A0/A1/A2/C12 and /A0/B1/B2/B3/B4/D1 1.2 Metadata under /A0/B1, but not including content under the directory /A0/B1/B2/B3/B4/D1 1.3 Metadata under directories /A0/A1/A2/C11 and /A0/A1/A2/C12 1.4 Metadata under directory /A0/B1/B2/B3/B4/D1

                       表3  匹配表示例Table 3 Example of matching table

(3)匹配表运行示例(3) Example of matching table operation

系统启动时解析保存匹配表的文件,读至以下三个字符串数组,常驻内存:When the system starts, it parses the file that saves the matching table, and reads the following three string arrays, which are resident in memory:

Sindex[i]:保存匹配表中第i个记录的索引值,匹配表中的索引值递增。它并非必需,但是可以简化和方便程序的编写;Sindex[i]: Save the index value of the i-th record in the matching table, and the index value in the matching table is incremented. It is not required, but can simplify and facilitate the writing of programs;

Spath[i]:保存第i个条目中字段“路径”的内容;Spath[i]: save the content of the field "path" in the i-th entry;

sDS[i]:保存第i个条目中字段“目录服务器名称”的内容。sDS[i]: Save the content of the field "Directory Server Name" in the i-th entry.

Sindex[i],spath[i],sDS[i]是相互对应的一套。从数组[1]开始记录文件。sindex[0]的值为表中存在的条目数;spath[0]为空;sDS[0]为空。Sindex[i], spath[i], sDS[i] are a set corresponding to each other. Start recording the file from array[1]. The value of sindex[0] is the number of entries existing in the table; spath[0] is empty; sDS[0] is empty.

收到用户请求访问的文件或目录路径path后,首先遍历数组spath,如果spath[i]的值为path,返回sDS[i];如果spath中不存在path,则返回path的父目录所在的目录服务器们,或者父父目录直至整个域的根目录。After receiving the file or directory path path requested by the user, first traverse the array spath, if the value of spath[i] is path, return sDS[i]; if path does not exist in spath, return the directory where the parent directory of path is located servers, or parent parent directories up to the root of the entire domain.

Claims (4)

1, a kind of metadata layer management method of memory virtualization system the steps include:
(1) whole memory virtualization system is divided into a plurality of domain logics, the metadata in each domain logic is distributed on a plurality of LIST SERVER;
(2) meta data server, the LIST SERVER that is used for managing this territory are set in each domain logic;
(3) matching list is set in each meta data server, this matching list comprises two fields at least: the root directory of the sub-logic tree of being preserved on LIST SERVER title and the LIST SERVER of metadata is used for the direct LIST SERVER that user's request is directed to its place;
(4) when the user files a request, from user's request, extract and to visit the path of catalogue;
(5) in the matching list in this territory, coupling is grown in the path, judged in matching list, whether there is the LIST SERVER that meets;
(6) if there is the LIST SERVER that meets, the LIST SERVER of directly visit coupling, and return results; Otherwise demonstration error message.
2, method according to claim 1 is characterized in that: also be provided with the catalogue copy in the described matching list.
3, method according to claim 1 is characterized in that: also be provided with index field in the described matching list.
4, a kind of metadata layer management system of memory virtualization system, this system is divided into a plurality of domain logics, metadata in each domain logic is distributed on a plurality of LIST SERVER, it is characterized in that: each domain logic specifies a station server as meta data server (2), specifies multiple servers as the LIST SERVER of preserving metadata information; Be provided with matching list (3) and matching table management module (5) in the meta data server (2), matching list (3) comprises the root directory of the sub-logic tree of being preserved on LIST SERVER title and the LIST SERVER of metadata at least, matching table management module (5) is used for that the record of matching list (3) is safeguarded, the parsing of matching list file and according to the route matching of user's request to suitable LIST SERVER or show error message.
CN 200310111436 2003-11-24 2003-11-24 Metadata hierarchy management method and system of storage virtualization system Expired - Fee Related CN1255748C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200310111436 CN1255748C (en) 2003-11-24 2003-11-24 Metadata hierarchy management method and system of storage virtualization system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200310111436 CN1255748C (en) 2003-11-24 2003-11-24 Metadata hierarchy management method and system of storage virtualization system

Publications (2)

Publication Number Publication Date
CN1545047A CN1545047A (en) 2004-11-10
CN1255748C true CN1255748C (en) 2006-05-10

Family

ID=34336101

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200310111436 Expired - Fee Related CN1255748C (en) 2003-11-24 2003-11-24 Metadata hierarchy management method and system of storage virtualization system

Country Status (1)

Country Link
CN (1) CN1255748C (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101826054B (en) * 2009-03-04 2011-12-07 安凯(广州)微电子技术有限公司 Memory management method of micromemory system
US10324903B1 (en) 2017-12-28 2019-06-18 Dropbox, Inc. Content management client synchronization service

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4808409B2 (en) * 2005-01-14 2011-11-02 株式会社日立製作所 Sensor network system, sensor data search method and program
KR100610904B1 (en) * 2005-03-03 2006-08-09 엘지전자 주식회사 Meta data parsing method for providing multimedia service and mobile communication terminal using same
JP2008078713A (en) * 2006-09-19 2008-04-03 Sony Corp Recording apparatus and method, program, and reproducing apparatus and method
CN100595761C (en) * 2007-12-29 2010-03-24 中国科学院计算技术研究所 A Metadata Management Method for Split Namespace
US8423581B2 (en) 2009-10-07 2013-04-16 International Business Machines Corporation Proxy support for special subtree entries in a directory information tree using attribute rules
CN101697168B (en) * 2009-10-22 2011-10-19 中国科学技术大学 Method and system for dynamically managing metadata of distributed file system
CN102024016B (en) * 2010-11-04 2013-03-13 曙光信息产业股份有限公司 Rapid data restoration method for distributed file system (DFS)
CN102982151B (en) * 2012-11-27 2015-04-01 南开大学 Method for merging multiple physical files into one logic file
WO2014101000A1 (en) * 2012-12-26 2014-07-03 华为技术有限公司 Metadata management method and system
CN103870588B (en) * 2014-03-27 2016-08-31 杭州朗和科技有限公司 A kind of method and device used in data base
CN106980697A (en) * 2017-04-07 2017-07-25 广东浪潮大数据研究有限公司 A kind of catalogue distribution querying method and device
CN120631889A (en) * 2024-03-12 2025-09-12 中兴通讯股份有限公司 Directory tree metadata management method, system, electronic device and storage medium

Cited By (40)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101826054B (en) * 2009-03-04 2011-12-07 安凯(广州)微电子技术有限公司 Memory management method of micromemory system
US10324903B1 (en) 2017-12-28 2019-06-18 Dropbox, Inc. Content management client synchronization service
US10599673B2 (en) 2017-12-28 2020-03-24 Dropbox, Inc. Content management client synchronization service
US10671638B2 (en) 2017-12-28 2020-06-02 Dropbox, Inc. Allocation and reassignment of unique identifiers for synchronization of content items
US10691720B2 (en) 2017-12-28 2020-06-23 Dropbox, Inc. Resynchronizing metadata in a content management system
US10726044B2 (en) 2017-12-28 2020-07-28 Dropbox, Inc. Atomic moves with lamport clocks in a content management system
US10733205B2 (en) 2017-12-28 2020-08-04 Dropbox, Inc. Violation resolution in client synchronization
US10762104B2 (en) 2017-12-28 2020-09-01 Dropbox, Inc. File journal interface for synchronizing content
US10776386B2 (en) 2017-12-28 2020-09-15 Dropbox, Inc. Content management client synchronization service
US10789269B2 (en) 2017-12-28 2020-09-29 Dropbox, Inc. Resynchronizing metadata in a content management system
US10866964B2 (en) 2017-12-28 2020-12-15 Dropbox, Inc. Updating a local tree for a client synchronization service
US10872098B2 (en) 2017-12-28 2020-12-22 Dropbox, Inc. Allocation and reassignment of unique identifiers for synchronization of content items
US10877993B2 (en) 2017-12-28 2020-12-29 Dropbox, Inc. Updating a local tree for a client synchronization service
US10922333B2 (en) 2017-12-28 2021-02-16 Dropbox, Inc. Efficient management of client synchronization updates
US10929427B2 (en) 2017-12-28 2021-02-23 Dropbox, Inc. Selective synchronization of content items in a content management system
US10936622B2 (en) 2017-12-28 2021-03-02 Dropbox, Inc. Storage interface for synchronizing content
US10949445B2 (en) 2017-12-28 2021-03-16 Dropbox, Inc. Content management client synchronization service
US11003685B2 (en) 2017-12-28 2021-05-11 Dropbox, Inc. Commit protocol for synchronizing content items
US11010402B2 (en) 2017-12-28 2021-05-18 Dropbox, Inc. Updating a remote tree for a client synchronization service
US11016991B2 (en) 2017-12-28 2021-05-25 Dropbox, Inc. Efficient filename storage and retrieval
US11048720B2 (en) 2017-12-28 2021-06-29 Dropbox, Inc. Efficiently propagating diff values
US11080297B2 (en) 2017-12-28 2021-08-03 Dropbox, Inc. Incremental client synchronization
US11120039B2 (en) 2017-12-28 2021-09-14 Dropbox, Inc. Updating a remote tree for a client synchronization service
US11176164B2 (en) 2017-12-28 2021-11-16 Dropbox, Inc. Transition to an organization directory
US11188559B2 (en) 2017-12-28 2021-11-30 Dropbox, Inc. Directory snapshots with searchable file paths
US11423048B2 (en) 2017-12-28 2022-08-23 Dropbox, Inc. Content management client synchronization service
US11429634B2 (en) 2017-12-28 2022-08-30 Dropbox, Inc. Storage interface for synchronizing content
US11461365B2 (en) 2017-12-28 2022-10-04 Dropbox, Inc. Atomic moves with lamport clocks in a content management system
US11475041B2 (en) 2017-12-28 2022-10-18 Dropbox, Inc. Resynchronizing metadata in a content management system
US11500897B2 (en) 2017-12-28 2022-11-15 Dropbox, Inc. Allocation and reassignment of unique identifiers for synchronization of content items
US11500899B2 (en) 2017-12-28 2022-11-15 Dropbox, Inc. Efficient management of client synchronization updates
US11514078B2 (en) 2017-12-28 2022-11-29 Dropbox, Inc. File journal interface for synchronizing content
US11657067B2 (en) 2017-12-28 2023-05-23 Dropbox Inc. Updating a remote tree for a client synchronization service
US11669544B2 (en) 2017-12-28 2023-06-06 Dropbox, Inc. Allocation and reassignment of unique identifiers for synchronization of content items
US11704336B2 (en) 2017-12-28 2023-07-18 Dropbox, Inc. Efficient filename storage and retrieval
US11782949B2 (en) 2017-12-28 2023-10-10 Dropbox, Inc. Violation resolution in client synchronization
US11836151B2 (en) 2017-12-28 2023-12-05 Dropbox, Inc. Synchronizing symbolic links
US12061623B2 (en) 2017-12-28 2024-08-13 Dropbox, Inc. Selective synchronization of content items in a content management system
US12135733B2 (en) 2017-12-28 2024-11-05 Dropbox, Inc. File journal interface for synchronizing content
US12169505B2 (en) 2017-12-28 2024-12-17 Dropbox, Inc. Updating a local tree for a client synchronization service

Also Published As

Publication number Publication date
CN1545047A (en) 2004-11-10

Similar Documents

Publication Publication Date Title
CN110825748B (en) High-performance and easily-expandable key value storage method by utilizing differentiated indexing mechanism
CN103020315B (en) A kind of mass small documents storage means based on master-salve distributed file system
US10114908B2 (en) Hybrid table implementation by using buffer pool as permanent in-memory storage for memory-resident data
CN106874383B (en) Decoupling distribution method of metadata of distributed file system
CN102158546B (en) Cluster file system and file service method thereof
US9405487B2 (en) Media aware distributed data layout
CN1255748C (en) Metadata hierarchy management method and system of storage virtualization system
US8543596B1 (en) Assigning blocks of a file of a distributed file system to processing units of a parallel database management system
CN103150394A (en) Distributed file system metadata management method facing to high-performance calculation
US20130013890A1 (en) Database system
CN105677826A (en) Resource management method for massive unstructured data
JP2015512604A (en) Cryptographic hash database
CN109213760B (en) High-load business storage and retrieval method for non-relational data storage
CN117687970B (en) Metadata retrieval method and device, electronic equipment and storage medium
CN103473337A (en) Massive catalogs and files oriented processing method in distributed type storage system
US20220342888A1 (en) Object tagging
CN117120998A (en) Method and apparatus for reading data stored in a tree data structure
CN1547137A (en) Mass file management system and method based on database
US7844596B2 (en) System and method for aiding file searching and file serving by indexing historical filenames and locations
Xu et al. Enhancing HDFS with a full-text search system for massive small files: W. Xu et al.
CN102169497A (en) Method and device for managing metadata through bitmaps
CN111581123B (en) Class-based locking of memory allocations
CN111274259A (en) Data updating method for storage nodes in distributed storage system
Zhang et al. PetaKV: building efficient key-value store for file system metadata on persistent memory
CN111831423A (en) A method and system for implementing Redis in-memory database on non-volatile memory

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C19 Lapse of patent right due to non-payment of the annual fee
CF01 Termination of patent right due to non-payment of annual fee