CN101419626B - An Application-Oriented File System Namespace Management Method - Google Patents

Abstract

The invention relates to an application-oriented file system namespace management method which belongs to the technical field of computer storage and aims at solving the problem that the file application semantics is not included in the metadata extension contents in the existing file systems so as to shorten the metadata access time and improve the application access efficiency. The method comprises three steps, i.e. extracting application semantics, cutting namespace and accessing cut namespace. By extracting the application semantics between the processes of the application program and the file accessed by the application program, the file system namespace that faces the processes of each application program is cut so that the metadata access time of the application program is shortened and the access efficiency is improved, and the method is suitable for the file system with more and more complex namespace.

Description

An Application-Oriented File System Namespace Management Method

technical field

The invention belongs to the technical field of computer storage, and in particular relates to an application-oriented file system name space management method.

Background technique

With the growth of the personal computer market and the continuous improvement of the data storage capacity of storage devices, the amount of personal data is becoming larger and larger. As the basis for managing user data, the file system's namespace becomes more and more complex, which leads to lower and lower file system metadata access performance.

In a computer system, a file includes two parts: metadata and data. Metadata is a description of attributes such as file length and type. The name of a file is an item in the metadata and is used to identify a file; a file system is a collection of files. , the name space of a file system is the collection of names of the files it includes.

The namespace accessed by the process of the application program is a collection of names of the files accessed by the process.

File system metadata plays an important role in the semantic file system, and many semantic information are expressed in the form of metadata. Current file systems do not pay enough attention to metadata, so they cannot make full use of file system metadata for content search, let alone provide some advanced semantic-based associative data access. At present, many semantic file systems extend file attributes in metadata to include various types of semantic information, and use these extended metadata to improve file retrieval efficiency.

At present, the content of metadata extension in semantic file system can roughly include the following categories:

1 Running status information. Running state information is needed in many areas, such as applications and compilers. Applications require short-term and long-term storage of state information, such as license information, user information, passwords, DNS or SMTP servers, ORB information, etc. The compiler also produces information about the program's data types.

2 data model. Database management systems, Windows registry, and interface repositories all maintain information about data structures and schemas, such as data types, indexes, constraints, relationships, and interfaces.

3 multimedia. Auditory and visual information like pictures, videos, music, etc. cannot be easily extracted from them in a useful format for query. Metadata information is therefore required to provide relevant access. Usually, this information may be entered after manually classifying data according to keywords or after extracting feature styles through image processing programs.

Although many current file systems, such as Microsoft's WinFS and Apple's Spotlight, provide some degree of semantic information, compared to a semantic file system built using a wide range of open data model methods, the semantic information they contain It's not enough. Most filesystems use a hierarchical directory structure, which is only suitable for filesystems with small or medium namespaces. In a large-scale data file system, the classification and retrieval of files will become very difficult. Users urgently need more effective methods for organizing files and retrieving files.

Contents of the invention

The invention proposes an application-oriented file system name space management method, which solves the problem that the extended content of metadata in the existing file system does not include the application semantics of the file, so as to shorten the metadata access time and improve the efficiency of application access.

An application-oriented file system name space management method of the present invention comprises the following steps:

(1) application semantics extraction step: run each application program in advance, track the access of the process of each application program to the file, and add a record metadata in the file to save the process name of the application program that has accessed the file; then According to all the record metadata, extract the application semantics, each application semantics describes the relationship between a process and the accessed file, and save each application semantics in the form of an XML file in the database, waiting to be called;

(2) Namespace clipping step: when this step is called, the corresponding application semantics are retrieved from the database according to the process name of the application, and the namespace associated with the process is clipped out according to the application semantics, and stored in the form of a DOM tree reside in memory, and then return;

(3) The access steps of clipping the name space: actually run the application program, when the process of the application program accesses the file, the file system redirects the received access request to the clipping name space associated with the process, if the redirection is successful, then Complete the access request; otherwise, call the namespace pruning step, and then send an access request to the file system.

The application-oriented file system name space management method is characterized in that:

(1) The step of extracting the application semantics, the following process is carried out in sequence:

(1.1) Start each application program, and when a file is accessed, proceed in sequence;

(1.2) Add the name of the access process to the record metadata of the accessed file F. The format of the record metadata is A:B:C:..., where F is the name of the file to be accessed, and A, B, and C are the access process respectively. The process name of the application for this file;

(1.3) All applications end, according to the record metadata of all accessed files, extract the application semantics, the format is A=F1:F2:F3:..., where A is the process name, F1, F2, F3 are the accessed file name;

(1.4) Write the application semantics of the process of each application program extracted into an XML file, save it in the database, and end;

(2) The step of clipping the name space, the following process is performed in sequence:

(2.1) When the step is called, obtain the ID parameter of the application process;

(2.2) according to the ID parameter of described process, access its data structure, obtain process name A;

(2.3) read out the XML file containing the application semantics of process name A from the database;

(2.4) The read XML file is parsed into a DOM tree, and the metadata corresponding to all file names included in the application semantics of the process name A is added to the DOM tree, and the DOM tree is resident in the memory to obtain the same as A associated clipped namespace, end;

(3) The access step of the clipping name space is to carry out the following process:

(3.1) When the process A of the application program accesses the file, it sends an access request to the file system and proceeds to the next step;

(3.2) The file system obtains the ID parameter of process A from the request packet, and then redirects the access request to the clipped name space associated with cA, and proceeds to the next step;

(3.3) The redirection operation checks whether the DOM tree corresponding to the ID parameter of process A exists in the memory, and if it exists, it will turn to the process (3.5), and if it does not exist, proceed to the next step;

(3.4) call the name space clipping step, obtain the name space of the clipping associated with process A, turn process (3.2);

(3.5) complete the file access request in the clipped name space associated with process A, and continue the application program;

(3.6) When the application finishes exiting, clear in memory the clipped namespaces associated with all processes it contains, end.

The present invention saves each application semantics in the database with an XML file by extracting the associated information between the process of the application program and the accessed file (i.e. the application semantics of the file), avoiding repeated extraction caused by restarting the system; and utilizing The application semantics of these files cuts the name space associated with the application process, fully considers and utilizes the huge difference in the name space of the system and the application process, shortens the time for the application process to access metadata, and improves the access to metadata Efficiency performance, suitable for file systems with increasingly complex namespaces.

Description of drawings

Fig. 1 is a schematic flow sheet of the present invention;

Fig. 2 is a schematic flow chart of the application semantic extraction step of the present invention;

Fig. 3 is a schematic flow chart of the name space clipping steps of the present invention;

FIG. 4 is a schematic flow diagram of the access steps of the clipping name space of the present invention;

Detailed ways

Assume that an application program consists of two processes, the process names are PA and PB respectively; a total of five files are accessed, namely File1, File2, File3, File4, and File5.

Fig. 2 is a schematic diagram of application semantic extraction steps;

(1) Start this application program, and when a file is accessed, proceed sequentially;

(2) Add the name of the access process in the record metadata of the accessed file;

(3) All application programs are finished, and the application semantics are extracted according to the record metadata of all accessed files, and the application semantics are PA=File1:File2:File3, PB=File4:File5;

(4) Write the application semantics of the extracted processes of each application program into an XML file and save it in the database;

(5) END.

Fig. 3 is a flow chart of name space clipping steps;

(1) When the step is called, obtain the ID parameter of the application process;

(2) according to the ID parameter of described process, access its data structure, obtain process name PA;

(3) read out the XML file containing the application semantics of the process name PA from the database;

(4) parse the read XML file into a DOM tree, add the metadata corresponding to File1, File2, and File3 included in the application semantics of the process name PA to the DOM tree, and reside the DOM tree in memory, Get the clipped namespace associated with the PA;

(5) END.

Fig. 4 is the flow chart of the access process of clipping name space;

(1) When the process PA of the application program accesses the file File1, it sends an access request to the file system, and proceeds to the next step;

(2) The file system obtains the ID parameter of the process PA from the request packet, and then redirects the access request to the clipped name space associated with the process PA, and proceeds to the next step;

(3) the redirection operation checks whether the DOM tree corresponding to the ID parameter of the process PA exists in the memory, if there is, then turn the process (5), and if it does not exist, proceed to the next step;

(4) call name space clipping step, obtain the name space of clipping associated with process PA, turn process (2);

(5) complete the access request of the file File1 in the clipped name space associated with the process PA, and continue the application program;

(6) When the application program finishes exiting, clear the clipped name space related to the process PA it includes in the memory;

(7) END.

Claims (2)

1. an application oriented file system name space management method comprises the following steps:

(1) application semantics extraction step: move each application program in advance, follow the tracks of of the visit of the process of each application program file, and increase hereof one the record metadata be used for preserving the process name of the application program of visiting this document; Then according to all record metadata, extract application semantics, every application semantics is described the relation between a process and the institute's access file, and the form of each application semantics with the XML file is saved in the database, and wait is called;

(2) name space cutting step, order is carried out following process:

(2.1) when this step is called, obtain the ID parameter of program process;

(2.2) according to the ID parameter of described process, visit its data structure, obtain process name A;

(2.3) from database, read the XML file of the application semantics that comprises process name A;

(2.4) the XML document analysis of reading is become dom tree, the All Files name metadata corresponding that the application semantics of process name A comprises is added in this dom tree, and this dom tree is resided in the internal memory, obtain the name space of the cutting related, return then with A;

(3) accessing step of cutting name space, carry out following process:

(3.1) during the process A access file of application program, send request of access, carry out next step to file system;

(3.2) file system obtains the ID parameter of process A from request package, request of access is redirected to the name space of the cutting related with process A again, carries out next step;

(3.3) redirect operation checks in internal memory whether the dom tree corresponding with the ID parameter of process A exists, and exists then to turn over journey (3.5), does not exist and then carries out next step;

(3.4) call name space cutting step, obtain the name space of the cutting related, turn over journey (3.2) with process A;

(3.5) in the name space of the cutting related, finish the request of access of file, proceed application program with process A;

(3.6) application program is finished when withdrawing from, and removes the name space of the cutting relevant with its all included processes in internal memory, finishes.

2. application oriented file system name space management method as claimed in claim 1 is characterized in that:

(1) described application semantics extraction step, order is carried out following process:

(1.1) start each application program, when file was accessed, order was carried out;

(1.2) title of interpolation visit process in the record metadata of accessed file F, the form of record metadata is A:B:C:..., and wherein F is accessed filename, and A, B, C are respectively the process name of the application program of this file of visit;

(1.3) all application programs finish, and according to the record metadata of all accessed files, extract application semantics, and its form is A=F1:F2:F3: ..., wherein A is a process name, F1, and F2, F3 are accessed filename;

(1.4) with the application semantics of the process of each application program of being extracted, write as the XML file, be saved in the database, finished.

Images