Distributed transaction Processing

Distributed | transaction processing Distributed transaction processing
Zhang Jianji 01-6-22 04:48:27

The US Sybase company released its beta version of PowerBuilder 6.0 this July, and a formal version will be launched in the near future, with support for distributed transactions being added to the new version The most powerful function. As early as 1995, powersoft the development strategy of "distributed transaction", and supported the function of Distributed transaction processing in PowerBuilder 5.0 released in 1996, Pen The issue of how to create a distributed PowerBuilder application has been discussed in previous lectures, but there is little feedback from readers. In the past, a programmer using PowerBuilder to develop client/server applications was less likely to think of using distributed transactions to segment applications, so the general user's overview of distributed transactions And the meaning is not very well understood, here we discuss the Distributed transaction processing and its application. Level Three architecture first we introduce the concept of level three architecture. The so-called level refers to a functional division, we have developed a database application software is generally based on client/server structure, we call the two-tier architecture. In other words, the whole system can be divided into two functional blocks, the first level includes the software application layer and the performance layer, resides in the client. We use PowerBuilder to develop applications primarily for the first level, running on the client. The second level contains the components of the database and the server. A SQL based database management system is generally installed on the server side, the application software on the server side of the operation is mainly data storage and retrieval. In level Two mode, there are some application logic stored in the form of procedures and triggers Stored on the server side to optimize the performance of the server, but most of the application logic is placed on the client. The level three model divides the system into three different "levels": performance level, business logic level, and data access level. The performance level is the function of processing the user interface, the data access level is the data source, in the usual situation refers to the database; The business logic level is the newly added level, which refers to the intelligent decision making in the program. Part of the function. In early applications, this part of the function is not very complex, generally put it on the performance level, and a small number of stored procedures or triggers in the form of data access level , and with the development of software engineering, more and more complex softwares, the most function in software is at this level. The function of an MIS system is from the previous simple query to a certain table, it develops to the classification statistic summation that involves multiple tables, according to the complex formula analysis and calculation, making decision support etc. , such enhancements are still being placed at the performance level, making the client more and more overwhelmed, so someone is proposing to separate the business logic out of the system , which forms a level three structure. With the further development of level three structure, the software that runs at commercial logic level is usually written as a special soft is different from the database server, which we call the application server. In a network, can have a number of different functions of the application server, for clients or other application servers to provide professional services, so that three-tier structure has become a N-class , this is the so-called distributed computing method. Advantages of distributed computing: using distribution Computing has many technical advantages, including: Logic encapsulation: This is the most attractive feature in distributed mode, which is based on what has been done by the client Part of the logic of the service is separated from the client. When the company needs to dynamically change the business logic rules for an application, simply change the application server's program without having to change the client user interface , which eliminates the need to interrupt users, release new interface software for end users or come in person to install and retrain users, and improve productivity. This multi-level model is useful for businesses that need to change applications frequently and quickly. Thin client: The most notable feature of this type of application at runtime is the reduction or elimination of the traditional two-tier architecture, client-centric or "fat customer" mode , lighten the client's function burden, make its swelling become "thin customer". "Fat customer" is the user is very distressed things, users for the use of more powerful software, you must pay high maintenance costs, and constantly for personal computer hardware and software equipment upgrades. The recent popular NC is also to see the average user in the maintenance of PC Run-time overload, and put forward through the network part of the task to the server to complete. The two methods have similarities. Performance: Performance improvement is the main reason why the three-tier model is ultimately used by users. The separation of complex application and business logic is handled by a dedicated application server, which can increase the speed of application execution and reduce the traffic of network calls. But there is a certain cost to this performance improvement. This is the time to develop Application logic is divided into client logic and server-side logic, which increases the complexity of the design. Security management: In distributed computing mode, because all business logic resides on the server side, information Management department can easily monitor the operation of the server , it is easy to control the number of access servers and the numbers of people who are dealing with the server application. This can greatly simplify administrator management of the system, reduce the workload of system maintenance, and ensure the reliable operation of the system. The broadest and most successful application of distributed transactions is the number internet/intranet technology, although many people do not realize that the Internet is a three-tier architecture representative. As the use of internet/intranet continues to deepen, the information queried by the Web server is not limited to the static hypertext files stored on the server side in a file format. At this point we need to use a relational database to store the changed data, and between the Web server and the database server with JDBC, CGI and other protocols to establish a connection between the two , enabling the Web server to implement an ad hoc query to the database and return the results to the browser. We can find that in such an intranet structure, the browser--web server--the database server corresponds to the client--Application server--database server level three architecture Three of the links, the Intranet is a typical application of the three-tier architecture. Therefore, we can think that the technical advantages of Internet mode, such as simple user interface, easy maintenance of managers and so on are also the advantages of multi-level computing methods. In addition, because PowerBuilder supports distributed transactions, which makes it easy for PowerBuilder to support the development of the Internet as a special case for distributed transactions---. Development of distributed applications although distributed computing is a newer concept, using PowerBuilder to develop distributed applications is very simple, as long as you use the PowerBuilder users from You can quickly grasp the development of distributed powerbuilder by defining objects that are not considered. In PowerBuilder, the functionality at one end of the application server is implemented through an unreachable user object, and developers can place PowerBuilder custom user objects on the application server One side, called a remote object, places the proxy object of the user object on the client. In addition, there is a new transport object on the server side that listens for requests from any user or other application server and has a connection object on the client to establish the same remote The connection to the object. The client's application connects to the application server through the connection object, and the client application can invoke the object function on the application server as if it were a local object when the connection is established. Of course, the user we refer to here can be a customer in the general sense, or an application server calling another application server as a customer. The business logic in the remote user object can be written in PowerScript, as a special case of a user object that can call any function or use a database command , it can also use an datastore to encapsulate database access, PowerBuilder 5.0 can support all but visual objectsThe data type. Using this form of call on the client is like passing a parameter over the network to the server as if using a stored procedure, and the server's running results are returned to the client.