Asynchronous replication of data changes by distributed update requests

Description Submit all comments and votes

BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention is to a method for propagating changes made to a database in a computer system to other databases in computer systems connected in a network. In particular, a method for synchronizing changes to relational databases is described.

2. Background Information

In many computer systems, the processing and storage components are distributed geographically and interconnected by means of one or more communication networks. Data is often distributed among the components and stored in relational databases. One such computer program for creating and managing relational databases is the DATABASE 2 (a trademark of IBM Corp.) program product, available commercially from IBM Corp. In large enterprises, it is often desirable for the databases in each computer, or node, in the network to contain identical information, such as address or phone data for employees.

The problem in such a distributed environment, however, is one of ensuring that any changes made to one database are propagated to the other databases in the system so that data remains consistent. This problem has been addressed in the prior art by automatically "pushing" any changes throughout the rest of the network. While this solution may be satisfactory under some conditions, it does not work well in complex networks. It is therefore desirable to provide a method for synchronizing changes to relational databases in a network.

OBJECTS OF THE INVENTION

It is the object of this invention to provide a method for synchronizing changes to relational databases in a computing network.

It is a further object of this invention to provide a method for synchronizing changes to databases in a peer to peer relationship.

It is still another object of this invention to provide a method for synchronizing changes to databases in a hierarchical relationship.

SUMMARY OF THE INVENTION

These objects, and others to be described, are accomplished by the following method in which the node containing the "changed" database is referred to as the "Collectee" and the database to be updated is referred to as the "Collector". Data variables that exist in databases D1 and D2 are said to be shadowed in D1 if updates occur in D2 but not D1.

As updates are made on the Collectee node, each record is timestamped with the date/time of the update. If a record is deleted, a physical deletion does not occur but instead a delete indicator is turned on in the record.

In the first step, a first program (TP1) in the Collector node initiates a conversation with a second program 2 (TP2) in the Collectee node. TP1 instructs TP2 to send all updates to the table of interest (shadowed table) since the last conversation. TP2, in response to the call by TP1, receives and answers queries to retrieve any changes made since the last update and sends the data to TP1. TP1 receives the data and updates the shadowed table in its own machine. Control tables contained in both machines are updated to indicate the last date and time of updating.

BRIEF DESCRIPTION OF THE DRAWINGS

FIGS. 1A-C shows schematic representations of three types of network structures.

FIG. 2 shows a schematic representation of a hierarchical type of network in a collecting node.

FIG. 3 shows representations of the shadowing support tables.

FIG. 4 shows a flow diagram for the Collector (TP1) and Collectee (TP2) programs.

FIG. 5 shows a pseudocode listing for the Collector program.

FIG. 6 shows a listing of the pseudocode for the Collectee program.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Present computer systems can be distributed in various forms. FIG. 1A shows a logical relationship in which processor 10 sits at the top and collects information from processors 12, 14, and 16. Processor 14 collects information from processors 18,20 and 22. In the hierarchical logical configuration, updates flow up the hierarchy to processor 10. These updates then eventually flow back down the hierarchy so that lower level processors may receive changes made by those processors at equal and higher levels. Processors 12, 14, and 16 collect updates from processor 10. Processors 18, 20, and 22 collect updates from processor 14.

FIG. 1B schematically illustrates what is known as a star configuration in which processor 30 is the central node and is connected to processors 32, 34, 36, 38, 40 and 42. FIG. 1C illustrates a peer network configuration in which each processor 50, 52, 54 and 56 has a connection to every other processor within the network. Network configurations are well-known in the computer industry and further discussion of network structures are beyond the scope of this description and unnecessary for an understanding of the present invention.

Regardless of the type of network, it is often necessary for all of the processors or nodes to contain identical information in their databases. In the preferred embodiment, the database to be considered is a phone directory/address book for a corporation. It is understood by those skilled in the art that the invention is extendable to all types of databases.

Since the invention for synchronizing the databases to be described herein is the same for all network structures, the detailed description will be limited to the hierarchical structure as further shown in FIG. 2. In this example, node 22 has recently been updated with changes to its phone directory/address book. It is referred to as the Collectee node. Node 14 is known as the Collector node because it collects data from the Collectee node 22. In turn node 14 is in the Collectee node for Collector node 10. The shadowing process is always initiated by the Collector node. This ensures that no undesired data is sent to a node. The Collector node can be any node within the network. A node doesn't need to be only a Collector, it can also be a Collectee in another shadowing process, so the place the node has within the network does not matter.

The network configurations shown represent logical data flows only. A line connecting two nodes only means that a data collection takes place between those two systems. Physically, there may be several other nodes in between the Collector and Collectee. As long as the Collector and Collectee can talk to each other, it doesn't matter what the physical configuration of the network is.

Referring to FIG. 3, the control table (shadow.sub.-- tbl) 60 for the database to be shadowed in the Collector node is illustrated. Shadow.sub.-- tbl 60 contains several data entries as follows:

key=identifier which uniquely identifies each row of unit data

XY . . . =represents columns in the data table

del? =logical indicator that record has been deleted

TLU=time last updated. (Time stamp when this row was last updated).

Also shown in FIG. 3 is the shadow control table 62 (Collectee.sub.-- tbl) which is contained in the Collectee node. This table contains the following data:

LUName=network address for Collectee node

TPName=program (TP2) to invoke on Collectee node program

PASSW=security of password of program TP2 on Collectee node

TLC=time last called. (A time stamp of the last time a successful conversation was completed normally with TP2).

DTC=delta time between collections (amount of time between collection calls to this node.)

TLS=time last serviced. (A time stamp of the last time a successful conversation was completed with TP1. Updated by TP2).

Referring now to FIG. 4, the method of the invention is as follows. The steps in the left side of the Figure take place in the Collector node and are implemented by TP1. The steps in the right side of the Figure take place in the Collectee node and are implemented by TP2. Of course, since a particular node can be both a Collector and Collectee at different times, each node contains both TP1 and TP2.

In Block 100, the Collector node checks the current time. Block 102 checks the TLC in the Collectee.sub.-- tbl 62 (FIG. 3). If the time since the last update exceeds the specified delta times (DTC) between conversational exchanges of the 2 nodes, then TP1 will initiate a conversation with TP2 in the Collectee node (Blocks 104, 106). If not, then Block 100 will continue, at regular intervals, to check the time until the delta time has been exceeded.

In the TP2 program in the Collectee node, initialization of the program is done in Block 200. Block 202 prepares a Structured Query Language data query to find data that has been changed in the phone/address book database. The changed data is obtained (Block 204) and sent to the Collector node (Block 206). When updating is complete, Block 210 updates the TLS in the Collectee.sub.-- tbl.

TP1 then receives the updated data, applies the updates to its phone directory/address book database (shadow.sub.-- tbl) (Block 108), and updates TLC in its Collectee.sub.-- tbl. FIGS. 5 and 6 contain pseudocode listings for the Collector and Collectee programs, also known as TP1 and TP2 respectively. While these examples employ the LU6.2 communications protocol, it is readily apparent that any suitable peer-to-peer communications protocol can be used.

While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that changes to the described method can easily be made without departing from the spirit and scope of the invention. For example, the computer network can be of any configuration and the database of any type. Accordingly, the invention shall be limited only as specified in the following claims.

* * * * *