Patent Application
20090077101
The present invention relates generally to databases and more specifically to handling database transactions in web based environments.
Sysplex (multiprocessor cluster-based) enabled infrastructures are utilized extensively for workflow of client applications. For example, they are utilized in automatic teller machine (ATM) transactions. It has been found that some banking customers have experienced significant operational issues when accessing data in such an infrastructure.
Any customer who has a sysplex-enabled infrastructure that uses a web application server, such as WebSphere, with a TCP/IP gateway (for example, IMS Connect) to access a database and sends their asynchronous transactions using dedicated persistent socket will receive a duplicate client ID problem. When the duplicate socket ID problem is generated, the transaction request that was sent will either hang, return an error message, or become lost. This is a severe problem because the customer's clients potentially will be left standing at an ATM machine waiting for a transaction that cannot be completed.
To describe these issues in more detail, refer now to the following description in conjunction with the accompanying figures. There are two scenarios in which the duplicate socket ID problem is generated.
In this scenario, when an ATM machine 20 whose client ID is ATM1 sends an asynchronous transaction request, the sysplex distributor 12 assigns the transaction request to a specific server, for example, server 14a. The resource adapter 19a inside server 14a sends the transaction request to the database system 17 through a TCP/IP gateway system 16 using a dedicated persistent socket connection 21a. The dedicated persistent socket connection 21 must have a unique client ID, which in this case is ATM1. The client ID identifies the client application and allows only transaction requests from that client application to reuse the socket connection; hence it is a dedicated persistent socket. The transaction is completed when the response is returned to the ATM machine 20 and the interaction ends. However, because the socket 21a is a dedicated persistent socket, that connection from the server 14a to the TCP/IP gateway 16 stays open to be reused for the next transaction request from the ATM1 client application.
If another transaction request is sent by the same ATM machine 20 (whose client ID is ATM1) and the sysplex distributor 12 assigns the request to another web application server, for example application server 14b, the same process occurs. The resource adapter 19b inside application server 14b sends the transaction to database system 17 through the TCP/IP gateway 16 using the persistent socket connection 21b that contains the ATM1 client ID. However, the server 14a already has the ATM1 client ID socket connection open.
The gateway 16 recognizes that the socket connection 21a is open on server 14a and rejects the transaction request from server 14n with the message that it is a duplicated client ID even though the interaction on server 14a had completed. Because each dedicated persistent socket connection 14 must have its own unique client ID and the ATM machine 20 only has one ID, a duplicate socket ID error is generated. If multiple transaction requests are sent through the ATM1 machine 20 and distributed to different servers by the SD 12, many transactions will be rejected due to the duplicate ID problem.
Another scenario that generates the duplicated client ID problem is if the customer uses one application server and creates multiple servant regions within that server. Similar to the SD, the server acts like a workflow manager and directs transaction requests to specific servant regions to increase server availability. The following figure illustrates the process flow in which the duplicated socket ID problem is generated.
This problem increases exponentially if any customer has a sysplex structure where there are multiple application servers and each of those servers have multiple servant regions.
To address this problem, a customer can choose to use shareable persistent socket for sending asynchronous transactions to the application server to bypass the duplicate client ID problem. However, with this method, the customers must rewrite their client applications and not specify a customer-specified client ID to an application server. This eliminates the capability of using the customer-specified client ID for tracking, security, and maintenance purposes.
Also, because the legacy application running in the database in most cases uses the customer-specified client ID to identify the client application for unsolicited messages, the legacy application must then be re-written for this solution. Furthermore, with this solution, using the sharable persistent socket increases the number of IMS output queues, thus impacting IMS storage requirements.
Accordingly, what is needed is a system and method that overcomes the above-described operational issues. The system and method should be cost effective, easily implemented and adaptable to web-enabled database systems. The present invention addresses such a need.
A system and computer readable medium for handling asynchronous database transactions in a web based environment is disclosed. The system and computer readable medium comprise providing a first ID from a device via a dedicated persistent connection and generating a second ID in an application server upon receipt of the first ID. The system and computer readable medium also include utilizing the first ID and second ID to obtain the appropriate data from a database system and to send the appropriate data to the device.
A system and computer readable medium in accordance with the present invention uses a database resource adapter, which runs inside an application server, to generate a socket ID internally for every input transaction running with the dedicated persistent socket connection and also saves the customer-specified client ID for database queue creation and legacy application. This operation is transparent to the customers' client application.
The present invention relates generally to databases and more specifically to handling database transactions in web based environments. The following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a patent application and its requirements. Various modifications to the preferred embodiments and the generic principles and features described herein will be readily apparent to those skilled in the art. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features described herein.
A system and computer readable medium in accordance with the present invention uses a database resource adapter, which runs inside an application server, to generate a socket ID internally for every input transaction running with the dedicated persistent socket connection and also saves the customer-specified client ID for database queue creation and legacy application. This operation is transparent to the customers' client application. The system and computer readable medium in accordance with the present invention has many advantages which are described herein below.
In this infrastructure 100, a transaction is sent from the ATM machine 20″. The ATM machine 20″ whose client ID is ATM1, sends the transaction through to a sysplex distributor 202 where the sysplex distributor 202 manages the workload and directs the transaction to an application server that is available, for example, application server 204a. The transaction is sent to the resource adapter 209a. The resource adapter 209a sends the transaction request to the database system 207 through a TCP/IP gateway 206 using a dedicated persistent socket connection 221a.
When the resource adapter 209a sends the transaction using a dedicated persistent socket connection 221a, the resource adapter 209a sends both the client ID and a generated socket ID in a message header to the TCP/IP gateway 206. In the message header, the generated socket ID is used to identify the socket name, for example, HWS001, and the client ID, ATM1, is passed as the queue name value inside the network message header.
The TCP/IP gateway 206 receives the transaction with the message header, reads the socket identifier flag only (as designed), and forwards the information to the database system 207 using the generated socket ID, HWS001, as the socket identifier. There is no change in the way the TCP/IP gateway 206 processes the transaction. The transaction is sent to a database queue 226a-226n that is identified by the queue name flag, ATM1. When the transaction is completed, the information is returned to the client application following the same route using the HWS001. The socket ID, HWS001, is then sent back to a connection pool of resource adapter 209a to be reused by the next transaction.
If the client applications send a second transaction the same process occurs. A transaction is sent from the ATM machine 20″, whose client ID is ATM1. The transaction is distributed to an available application server such as application server 204b. The resource adapter 209b sends the transaction using a dedicated persistent socket 221b, where the resource adapter 209b sends both the client ID and a generated socket ID in a message header. The resource adapter 209b uses a new randomly generated socket ID, which in this case is HWS00A. In the network message header, HWS00A is used to identify the socket name and the client ID, ATM1, is passed as the queue name value.
Similarly, the TCP/IP gateway 206 receives the data, reads the socket flag, and uses the generated socket ID HWS00A to identify the socket connection to the database system 207. The transaction is processed in the database queue 226a-226n that was identified by the queue name, ATM1. All transactions sent from the client ATM1 are processed on the same database queue 226a-226n. This allows the customers to track their transactions. The transaction is processed and the information is sent back to the client application following the same route.
A system and computer readable medium in accordance with the present invention uses the network message header to store customer-specified client ID so that the TCP/IP gateway 206 or follow-on middleware component has no knowledge of the special saved information. The information is saved in the message header for back-end database processing only.
There is no TCP/IP gateway dependency. Because a new socket ID is generated by the resource adapter 204a-204n internally and thus, bypasses any middle components, the TCP/IP gateway 206 does not need to be modified or have source code changes. This helps customers who use different TCP/IP gateways other than TCP/IP gateway 206.
A system and computer readable medium in accordance with the present invention creates only one database transaction output queue 226b for each client application. Rather than having multiple output queues for each new generated socket ID, only one queue is created by using the original customer-specified client ID name as the queue name. This greatly reduces the storage requirement for implementing and creating database system queues.
A system and computer readable medium in accordance with the present invention solves the duplicate client ID problem reported by the customers running application servers in the sysplex distributor environment. It also solves the problem for customers running application servers in the distributed environment.
A system and computer readable medium in accordance with the present invention requires no code change from the customer. The customers do not need to change their client application or legacy database application.
As described earlier, the duplicate client ID problem can happen in any sysplex-enabled environment for a web client application; and is problematic for any database customer who has this type of software and hardware architecture. A system and computer readable medium in accordance with the present invention avoids any changes from the customers' end and requires no TCP/IP gateway dependency.
With a system and computer readable medium in accordance with the present invention, when the resource adapter 209a-209n, which is inside an application server 204a-204n or any application server servant region, sends a transaction through TCP/IP gateway 206 to a database server, it also sends a message header with specific transaction data that contains a newly generated socket ID value as well as the customer-specified client ID value. The database resource adapter 209a-209n invokes a random number generator to uniquely create a generated socket ID for the input transaction so that every new input transaction will have a new ID created. This uniquely generated socket ID is used for identifying the socket connection 221a-221n, and it is stored in the TCP/IP gateway defined client ID section of the message header. As a result, the dedicated persistent socket always has a unique ID name for communicating with the TCP/IP gateway 206.
In addition to the uniquely generated socket ID, the customer-specified client ID is also passed to the TCP/IP gateway 206. However, the customer-specified client ID is passed to the database system 207 using an existing section of the network message header. This existing section of the network message header is not used by the TCP/IP gateway 206 and is a section reserved for the resource adapter 204a-204n only.
By passing the customer-specified client ID in a pre-existing reserved section of the message header, the customer-specified client ID is transparent to the TCP/IP gateway 206 or any follow-on middleware components. This eliminates any interface changes to or from the TCP/IP gateway 206. When the database system 207 receives the transaction with the socket ID (which is the generated socket ID) and the customer-specified client ID in the network message header, the database system 207 detects the flag set by the resource adapter 204a-204n and correctly uses the customer-specified client ID to create a database output queue 226a-226n for transaction processing. This operation in the database system 207 is important so that one and only one database output queue 226a-226n is created for the client application using the customer-specified client ID.
The database system 207 does not use the socket ID, which is the generated socket ID, to create a processing queue. However, the input socket ID is saved for output delivery. When the database system 207 delivers the transaction output to the TCP/IP gateway 207, the database system 207 uses the saved socket ID to communicate with TCP/IP gateway 206 for returning the output. With this unique handling of the code in the database system 207, the creation of multiple database transaction queues 226a-226n using the different socket IDs is avoided and eliminates any impact to the database system 207 storage requirements.
The system and computer readable medium in accordance with the present invention also applies to sysplex-enabled environments where one application server 204a-204n has several servant regions with multiple resource adapters (not shown). In all cases, no duplicate socket ID error is generated because the resource adapter 204a-204n sends a unique client ID for the socket identifier each time. The original client ID is stored in the network message header and used later by the database system 207 to contain all transactions in the same database output queue 226a-226n.
Although the present invention has been described in accordance with the embodiments shown, one of ordinary skill in the art will readily recognize that there could be variations to the embodiments and those variations would be within the spirit and scope of the present invention. Accordingly, many modifications may be made by one of ordinary skill in the art without departing from the spirit and scope of the appended claims.
Under 35 USC §120, this application is a continuation application and claims the benefit of priority to U.S. patent application Ser. No. 11/226,812 filed Sep. 13, 2005, entitled “METHOD AND SYSTEM FOR HANDLING ASYNCHRONOUS DATABASE TRANSACTIONS IN A WEB BASED ENVIRONMENT”, all of which is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11226812 | Sep 2005 | US |
| Child | 12212248 | US |
