Distributed transaction processing Zhang Jianzi 01-6-22 04:48:27
The US Sybase Division issued the Powerbuilder 6.0 Beta version in July this year, and the official version will be launched for a long time, and the support of sub-format processing is the new version of the most powerful function. As early as 1995, PowerSoft proposed a development strategy in the "distributed transaction" and supported distributed transaction processing in PowerBuilder 5.0 issued in 1996. The author also discussed related to the previous lecture. How to create a distributed PowerBuilder application, but it is very little feedback from reader. In the past, a programmer using PowerBuilder developed client / server application is less thought of using distributed transactions to split applications, so general users don't know much about the concept and meaning of distributed transactions, here we come further discuss Distributed transaction processing and its application.
Three-level system
First, we will introduce the concept of three-level system. The predicate refers to a functional division, and our application software is based on the cousin / server structure, which is called two-level system. That is to say, the whole system can be divided into two functional blocks, and the first level includes the application layer and the appearance layer of the software, resides in the customer. We use PowerBuilder to use the application mainly to use the first level to run the customer. The second level contains the components of the data library and servicers. A SQL-based data library pipe system is equally installed in the server side, and the application software should be a data storage and retrieval. There will be some application logic in the two-stage model to store the process of the program and the contact of the truncher to optimize the performance of the server, but most of the most applicable logies is placed at the client.
The three-level mode is divided into three different "grades": table, business logic and data access level. The table is the function of the processing user interface; the data access level is the data source, refers to the data library in the normal condition; the business logic is the newly added level, and the intelligence decision policy is made in the fingertips. In early applications, this part of the function is not very complicated, and it is generally placed in the performance level, and there is a small number of stored procedures or triggers in the form of data accesses, and with the development of software engineering , The software is increasing, and the most functional increase in software is at this stage. The function of a MIS system is developed by the earlier query on a table, and the classification statistics involving multiple tables. According to complex formula analysis, decision support, etc., such as putting these enhanced functions is still all The performance level will make the client increasingly overwhelmed, so someone has proposed that commercial logic is separated in the system, and the level is formed alone, which forms a three-level structure. With the further development of the three-level structure, it is generally written to the software running in business logic grade to become a dedicated software that is called to the client, and can complete a certain logical function, which is different from the database server. We call it application server. In a network, there can be a number of different types of active applications, which allows the client or its application server to provide a professional service. This, the three-level structure is developed into N-class, which is the score-based calculation method. Advantage of the distribution calculation:
It is a multi-faceted technical advantage, including:
Logic seal: This is the most fascinated feature of the layered mode. The root of this mode is to distinguish from the client with a part of the transaction logic completed by the guest machine. When the company needs to dynamically change a commercial logic rule of an application, just change the program of an application server, without changing the client user interface, there is no need to interrupt the user, re-issue new interface software for the end user or In order to install and rebuild users in the door, we have improved work efficiency. This multi-stage model is very helpful to the industry that needs to be often changed, quickly changed the application. Thin client: This type of application is the most significant feature of running at runtime, which is reduced to eliminate the traditional two-level architecture, the client-centric or "fertilizer customer" mode, reducing the client's function Burden, make it swelling into "thin customers". "Fertilizer" is a very annoyed thing that is used by households. It is necessary to pay high-level maintenance fees for the use of more powerful software, which must be upgraded for the hardware and hardware of the individual. NC in recent days is also to see that a general user is overwrick when maintaining the PC operation, and puts out a part of a part of the task to the servers. These two methods have a phase. Performance: The improvement of sexual energy is the main reason why the three-level model is most used. The complex application and business logic is separated from a special application server, which can be used to improve the accepted speed of the network, and the network-called communication. It is not possible to improve this performance. This is the time to send the application logic to the client logic and server-side logic, which increases the complexity of the design.
Security Management: In distributed computing mode, since all business logic resides in the server side, the information management department can easily monitor the operation of the server, and easy to control the access server and deal with the server application. Quantity. This can take advantage of the management of the system, reduce the work of the system, and ensure the reliable operation of the system.
The most widely used general and most successful application is used as an Internet / intranet technology. Although many people don't mean that the Internet is a representative of a three-level system. With the Internet / Intranet, the user's information on the Web server is not only limited to the static super textual file stored in the server side. At this time, we need to store changes with the relational database, and establish both connections between the Web Server and the Database Server Protocols such as JDBC, CGI, so that the web server can implement the sub-query of the database and will result Return to the browser. We can find that in such an intranet structure, the browser -Web server-database server corresponds to the three links in the client-application server-database server three-level architecture, and intranet is a typical Application of Three - level Architecture. Therefore, we can recognize that the technological advantages of Internet models such as user interface is simple, and managers are easy to maintain, etc., is also the advantages of multi-level accounting. In addition, the PowerBuilder is supported by PowerBuilder, which makes PowerBuilder easy to support a special example of the distribution-based affairs - the development of the Internet. Distribution
Although it is a new concept, but it is easy to use PowerBuilder to open the layout, but you will use the user's own unconventional inevisuality of users in PowerBuilder, you will quickly hold the discharge of the distribution PowerBuilder. In PowerBuilder, the functionality of the application server is implemented by unison user objects, the developer can put the object of the PowerBuilder's custom user on the application server side, referred to as a remote object, placing the user on the client. A proxy object of an object. In addition, there is a new TRANSPORT object on the side of the server to monitor any one of the users or it applies the server, and there is a connection pair in the client to build a connection with the distal object. The customer's application is connected to the connection server, and after connecting, the access to the client can be used to call the log in the objective function of the application server. Of course, users here can be a conventional customer, or it can be used to use a servant to call it to be used in one customer.
