KR20040055502A - Method for looking in dispersion conduct environment by using chord system - Google Patents

Abstract

PURPOSE: A lookup method on a distributed processing environment using a chord system is provided to perform lookup efficiently and fast on a P2P(Peer-to-Peer) environment that the information of resources is not collected in one place, the resources of a large scale are connected, and free registration/secession is possible. CONSTITUTION: To search the information of a certain number, the information is distributed/registered to the lookup lists of a number in the exponents of 2(301-3). In a distributed registration state, the information is sequentially searched while an appointed point address ID and a node ID of the client are obtained(302-1). The information is distributed/registered by using a list appointed point ID.

Description

Search Method in Distributed Processing Environment using Curd System {METHOD FOR LOOKING IN DISPERSION CONDUCT ENVIRONMENT BY USING CHORD SYSTEM}

The present invention relates to a retrieval method in a distributed processing environment using a curd system, and in particular, a peer-to-peer that is capable of joining and unsubscribing and connecting large resources without resource information gathering in one place. It's about how to get faster searches in your environment more efficiently.

In general, lookup in a network environment is one of the most important basic functions when using or implementing an application service.

In other words, it is important to ensure consistent search results in an environment where large users are connected, free to join and leave, and no network segmentation is foreseen. Consistent search results are those found and returned based on the exact status of the resource at the time of the search request, in case of network segmentation or subscription and withdrawal.

In addition, when requesting retrieval or modification from multiple nodes, the number of messages is reduced, and the supplementary use of a separate protocol that supports atomic communication and group communication that guarantees the order of messages according to transmission delay is useful. Results in a higher value.

In more detail, as the dissemination and use of computers become more common, the rapid increase in information and communication networks for acquiring and providing information through computers, and the rapid communication of information, and the explosive increase in the number of connected information devices and subscribers, Their names, addresses and various information needed for services have become complicated and vast.

There is a need for a system and service that can manage such information efficiently and systematically. In response to these demands, a Lightweight Directory Access Protocol (LDAP) directory is required. This LDAP directory is a set of rules and information that can be used to build a database that meets specific rules for information services for telecommunications services. Based on this, it defines various functions and services that can handle user's request.

In other words, an LDAP directory system means provision of a directory service to a user by building rules and services on a specific system. Directory clients handle the various directory services provided by the directory system, such as search, insert, delete, and update, on your behalf.

FIG. 1 processes a client's request for information by accessing a DB of another LDAP directory server connected through a communication means with a database in a conventional LDAP directory server, and the LDAP client server responds to the client. It is a block diagram for performing a method to make.

That is, referring to Fig. 1, a plurality of clients S1-1 to S1-n, an LDAP directory server 1 and a database (DB) 2 are included.

When any of the clients S1-1 to S1-n requests any information processing from the remote directory to the LDAP directory server 1 via the Internet S2, the LDAP directory server 1 Accesses the DB 2, processes the information requested by any client S1-1, and provides the processing result to any client S1-1.

In more detail, when any of the clients S1-1 to S1-n requests the LDAP directory server 1 for information, the LDAP directory server 1 connects to the LDAP connection ( One connection is opened and returned to an arbitrary client S1-1.

Any client S1-1 performs an LDAP operation process while connected to the LDAP directory server 1. At this time, an API such as adding LDAP data is called to obtain a result value corresponding to the requested information through the DB 2.

Any client S1-1 performs TCP / IP communication with the LDAP directory server 1 to find a desired name, and sequentially finds a naming list to communicate with the desired client.

In this way, each of the clients S1-1 to S1-n needs to find a desired name sequentially from the list of the LDAP directory server 1, thus targeting hundreds of thousands to millions of clients, such as a grid system with many targets looking up. In this case, there is a problem in that an increase in the number of searches used for searching and an economic burden due to this increase.

As such, a search technology in a distributed processing environment includes a "lightweight directory access protocol (LDAP) server system and operation method having information compatibility with a database management system" registered on March 14, 2000, 10-2000-0012832. "Communication Protocols for Directory Clients for Web Browsers and Their Structure," registered December 19, 1997, 10-1997-0070297, and "Lightweighted Directory Access, filed September 9, 1999, 10-1999-0038294. Transaction processing method in a protocol ".

As described above, the disclosed prior art will be described in detail. The LDAP server system and operation method having information compatibility with the database management system are commercially available in which the LDAP server is incompatible with the operating system and the language used in processing the information requested by the client. Technology that provides operating method and system technology that can access and use information data stored in database management system (DBMS).

Next, the communication protocol for the directory client for the web browser and its structure is a Java applet using JAVA APPLET. The LDAP and the directory server are provided through the LDAP client daemon. To send and receive data between the web browser directory interface and the LDAP client daemon, send the byte number of data to be sent first, and then send the byte number The data is transmitted, and the receiving end first checks the received data number and the received data number to complete data reception in the same case, and generates an error message when an incorrect value is received.

Next, the transaction processing method in the lightweight directory access protocol adds a transaction processing part to the LDAP structure, and adds a transaction API to reduce the weight so that a series of operations can be bundled and processed in the directory access protocol. A technology for providing a computer readable recording medium having recorded thereon a transaction processing method and a program for realizing the above-described method.

As described above, when looking at the technology disclosed in the prior patent, each client needs to find a desired name sequentially from the list of servers, so that the target of lookups is to target hundreds of thousands to millions of clients such as many grid systems. There remains a problem that increases the number of searches used to search and the resulting overall cost increases.

Accordingly, the present invention has been made to solve the above-mentioned problems, the object of which is that the information of the resources are not gathered in one place, can be joined and withdrawal arbitrarily and the peer-to-peer is connected to a large resource An object of the present invention is to provide a search method in a distributed processing environment using a curd system that enables faster and more efficient search in an environment.

In the present invention for achieving the above object, a search method in a distributed processing environment using a curd system is to distribute and register the corresponding information in the search list of the multiplier of 2 for any number to search for any number of information step; In the state of being distributedly registered, the step of sequentially searching for the corresponding information while obtaining the designated point address number (key_id) and the client itself number (node_id).

1 is a block diagram for performing a search method in a conventional distributed processing environment.

2 is a block diagram for performing a search method in a distributed processing environment using a curd system according to the present invention;

3 is a detailed flowchart illustrating a search method in a distributed processing environment using a curd system according to the present invention.

<Description of the symbols for the main parts of the drawings>

100: curd directory server S1-1 to S1-n: client

S2: Internet S3-1 to S3-n, 100-1: Search list

S4-1 to S4-n, 100-1-1: Replication list

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

2 is a block diagram for performing a search method in a distributed processing environment using a curd system according to the present invention, and includes a plurality of clients (S1-1 through S1-n) and a plurality of search lists (S3-1 through 1). S3-n, 100-1, a chord directory server 100, and replication lists S4-1 to S4-n, 100-1-1.

The plurality of clients S1-1 to S1-n and the curd directory server 100 perform bidirectional communication via the Internet S2, respectively, and the plurality of clients S1-1 to S1-n and the curd directory server are each. A plurality of search lists S3-1 to S3-n and 100-1 are connected to 100, respectively, and a replication list S4 to the plurality of search lists S3-1 to S3-n and 100-1. -1 to S4-n, 100-1-1) are respectively connected.

In such a state, when searching for 1024 pieces of information, it is necessary to first register only for 10 pieces of search list.

That is, the procedure of registering to each client search list (S3-1 to S3-n, 100-1) performs registration by using a key having a list designation point address, and a plurality of search lists (S3-1 to S3). Since the number of -n, 100-1) is n's power of 2 in each of the clients S1-1 through S1-n, n = 10, since 10's power of 2 is 1024. will be. At this time, since the curd directory server 100 is not necessary, a problem does not occur when the curd directory server 100 fails in such a structure.

With reference to the flowchart of FIG. 3, the searching method in the distributed processing environment using the curd system which concerns on this invention is demonstrated in more detail based on the above-mentioned structure.

First, if only eight pieces of information are to be searched instead of 1024 pieces of information, the information needs to be searched while being registered in only three search lists.

That is, in order to search for eight pieces of information, only three search lists need to be registered. For example, the registration process includes three search lists (S3-1 to S3-n, 100-1). Information corresponding only to S3-1, S3-2, and S3-3) is registered, and the procedure performs distributed registration using a key having a list designation point address (step 301).

That is, in the process of distributed registration, the search list S3-1 is connected to the client S1-1, and the distribution list S4-1 is connected to the search list S3-1. It may be possible (step 301-1).

The search list S3-1 may also be distributedly registered while the search list S3-2 having the next designated point address is connected, and the replication list S4-2 may be included in the search list S3-2. May be distributedly registered in a state in which () is connected (step 301-2).

The search list S3-2 may also be distributedly registered while the search list S3-3 having the next designated point address is connected, and the replication list S4-3 may be included in the search list S3-3. May be distributedly registered in a connected state (step 301-3).

As described above, in the state of being connected and distributed and registered, the corresponding information is retrieved while obtaining the designated point address number key_id and the corresponding client self number node_id (step 302).

That is, the designated point address number key_id and the corresponding client self number node_id are obtained using the key search function SHA-1 () (e.g., a hash function) (step 302-1).

Subsequently, a search list S3-1 connected to the designated point address number key_id and the corresponding client self number node_id, that is, the client S1-1 of FIG. 1 is obtained to determine whether the corresponding information exists (step S1). 302-2).

If the corresponding information exists in the determination step 302-2, the corresponding information is retrieved and the corresponding client self number (node_id) information is returned (step 302-3).

If the corresponding information does not exist in the determination step 302-2, a search list S3- connected to the next designated point address number key_id and the corresponding client self number node_id, that is, the client S1-2 of FIG. In step 2), it is determined whether the corresponding information exists (step 302-4).

If the corresponding information exists in the determination step 302-4, the corresponding information is retrieved and the corresponding client self number (node_id) information is returned (step 302-3).

If the corresponding information does not exist in the determination step 302-4, a search list S3- connected to the next designated point address number key_id and the corresponding client self number node_id, that is, the client S1-3 of FIG. In step 3), it is determined whether the corresponding information exists (step 302-5).

If the information exists in the determination step 302-6, the information is retrieved and the corresponding client self number (node_id) information is returned (step 302-3).

If the corresponding information does not exist in the determination step 302-6, it is considered that the search list has failed, and it is determined whether the corresponding information exists in the connected replication list S4-3 (step 302-6).

If the information exists in the determination step 302-8, the information is retrieved, and the corresponding client self number (node_id) information is returned (step 302-3).

If the corresponding information does not exist in the determination step 302-8, an error message is notified as if the corresponding search list and replication list are dead (step 302-7).

As described above, the present invention provides a fast and efficient search in a peer-to-peer environment in which a large amount of resources can be joined and unsubscribed and large resources are connected without the information of resources gathered in one place. In order to operate the LDAP protocol normally in a grid environment where users are connected, it can effectively cope with network partitioning and simultaneous joining and leaving of large nodes.

Claims (6)

In a search method in a distributed processing system environment, Distributing and registering information corresponding to a multiplier number search list of two for the arbitrary number to retrieve any number information; In the distributed registration state, obtaining the designated point address number (key_id) and the client itself number (node_id) sequentially searching for the corresponding information in the distributed processing environment using the system How to search. The method of claim 1, The information distributed registration is a search method in a distributed processing environment using a curd system, characterized in that distributed registration is performed using a key having a list designation point address. The method of claim 2, In the information distribution registration process, a search list S3-1 is connected to an arbitrary client S1-1 of a plurality of clients S1-1 to S1-n, and is replicated to the search list S3-1. Performing distributed registration in a state in which the list S4-1 is connected; The search list S3-1 is distributedly registered while a search list S3-2 having a next designated point address is connected, and a replication list S4-2 is connected to the search list S3-2. Registering in a distributed state; The search list S3-2 is distributedly registered while a search list S3-3 having a next designated point address is connected, and a replication list S4-3 is connected to the search list S3-3. And registering in a distributed state in the distributed state. The method of claim 1, wherein the searching for the corresponding information sequentially comprises: Obtaining a search list S3-1 connected to the client S1-1 corresponding to the designated point address number key_id and the corresponding client node number node_id, and determining whether the corresponding information is present; ; If the information does not exist in the first determination step, the information is included in the search list S3-2 connected to the client S1-2 corresponding to the next designated point address number key_id and the client self number node_id. Secondly determining whether is present; If the information does not exist in the second determination step, the information is included in the search list S3-3 connected to the client S1-3 corresponding to the next designated point address number key_id and the corresponding client node number node_id. Third determining whether is present; If the corresponding information does not exist in the third determining step, determining that the search list has failed and determining whether the corresponding information exists in the connected replication list; And if the corresponding information does not exist in the fourth determination step, notifying the error message as a case in which the corresponding search list and the replication list are dead. The method of claim 4, wherein If the corresponding information is present in the first, second, third, and fourth determination steps, after searching for the information, the method further includes the step of returning the corresponding client node number (node_id) information. Search method in processing environment. The method of claim 1, A search method in a distributed processing environment using a cud system, wherein the corresponding information can be searched even if a cud directory server in the cud system is not linked.