Parallel world: Java and XML Why do you have successful huqile (Reposted) keyword Java and XML home http://www-900.ibm.com/developerWorks/cn/java/parallel/index.shtml Simon Phipps Chief XML and Java Mission , IBM1999 Content: A new world mutual dependent new basic technology to users Parallel world Java: public property? Conclusion Reference Information About the author In the past few years, the calculation has gradually moved away from the original technology, and most of the most recent solutions have been determined to have a total cost (TCO) solution. But what constitutes TCO? This article discusses how interconnected complexity in a typical computer system affects TCO. And this interconnect complexity is Java technology and XML can be handled. What constitutes the overall cost cost (TCO)? It is difficult to say that everyone has different answers. It is often dependent on the most convenient way they find the most convenient solution. Most people agree that TCO is not just the sum of the price of each part of the system. Initially, the last most cost is from the cost of the system in the support environment. A popular reduction of TCO is to try to centrally manage independent systems, customer desktops, or both, but this is just part of the answer. It is best to minimize traffic, but actually cause management? Of course, the answer is changed. But it is not it itself. Isolated changes will only affect the change itself. We all know that a part of the change system will lead to support needs throughout the system. Ordinary computer systems typically lead to "entropy death", that is, cost exceeding the expected value, and orderly simpleness will become interconnected complexity. The method of cure this symptom may be centralized management, and the actual disadvantage will avoid dependent complex networks in the primary position. Java and XML avoid this happening by helping to eliminate automatic mutual dependence between systems, software, and data. Most of the new world supports and managed demands come from the software on the computer with a dependent network. To re-obtain simpleness, we need to remove dependencies. Where is the dependencies exist? There are several categories: Software to the platform software on the software platform to lift these dependencies to the platform is not easy, but the new world that gradually developed in the past decade is increasingly maturing and making it possible. Let us first consider the calculation model that is already enduring. It is easy to make a choice when the calculation is in the start phase. I can get any limited range of computers, write software running on this computer, and create file formats for storing data. Trouble is that software and data can only work on such a computer. When using another computer, another software must be used, or when using another software on the same computer, the same data cannot be used, and You must understand the new user interface. Through two standardization steps can solve many problems: Many people start using IBM PC, initially using DOS, then use Microsoft Windows. A certain degree of simplicity is back. But over time, it is increasingly clear that many of the scope of complexity is still mixed. In particular, the recognition of the platform does not break the platform dependence; this is precisely it is entirely dependent on each other. Therefore, when the update occurs, everything may break! In addition, the monopoly power of the data world is not standardized. As software depends on a specific level of platform, data is also related to specific levels of special brand software. So we will intend to have dependent complex networks, which can cause unstable changes, may also cause the collapse of the entire network. The top enemy of mutual-dependent calculation is unintentional. When building a computer solution, they all relate to the relationship between software, hardware, platform, and development tools. They are connected by visible connection clues with mutually dependent connection.
Over time, the cost of any solution is proportional to the amount of dependence between the supported parts. But because there is a lot of mutual dependence, the cost will grow in an index level, not linear growth. As a result, more additional costs of more interdependence elements may increase the lifetime cost. This starting point of uncomfortable growth is called a sprint point, and the situation above the sprint point is called entropy death. Before the sprint point, the system has been unintentionally dependent on the other part by selecting the system principle with mutual dependence, and it may be a solid foundation for entropy death (may be caused by other elements). The most common unintentional interdependence exists between the software and its claims. This is not to say that you should avoid all mutual dependencies; some mutual dependence is inevitable. However, in modern system specifications and design, you should use the same method as other cost drive factors to identify and adjust them. Usually, the software needs to be separated from the environment using it. In some cases, using native interface and binary is inevitable, but in these cases, "encapsulars" unrelated to the platforms of the machine code are almost always valuable. Figure 1. Cost vs. Nodes, for example, suppose a company uses a macro language for office suits as the foundation of office automation systems. One day, the company's IT team installed another software and uninformed a DLL file used by the office suite. They found that there is a macro that cannot be used. After a lot of work, they manage this macro again, but the new version requires an updated version of the spreadsheet program. In order to use the program, they have to install the new level of the office suite, and all the macros will not work in that future! Next, they debug all macros one by one, update and fix them. In other parts involved in these fixes, they find new versions of a database driver. Sadly, you need to use the latest version of the database. So they upgraded the database, and ..., hey, you can guess the rest. The new basic problem is caused by the transfer of the impact from the subsystem to subsystems. Most systems currently used the integrated computing basis can act as transport media, which allows changes to changes in other places. How to avoid this trap? The most critical is to cut off the connection between the data and the platform, and all of these use standard-based options so that the impact of version changes may be minimized. To reach this, we will isolate the change with the transfer medium (underlying platform) and prevent the impact of changes in the impact; we add the quarantine mentioned above. So what is the ideal standard basis? In the technical field shown in Figure 2, this basis should be covered: connecting the system to a network protocol that provides access to a network protocol that needs to be used to create a programming model solution for the solution. The data structure model of information allows for the appropriate user to access the appropriate data and solutions. Figure 2. Technical fields have changed the changes in computer circles in decade, and develop them into standards in the model. The following mappings are shown in Figure 3: Network: TCP / IP is now so broad for TCP / IP, so that it is no longer talking the subject. Pass: Web model stateless client / server stateless client / server computing is the transfer mechanism selected by many commercial computer users, and the number of users used continues to grow. Compared to the creation full state client, it does not require expensive maintenance and support costs, and the status is not maintained on the server, but most of the status "loans" is given.
Procedure: Java and JavaBeansjava have only four years since the release, but it has established itself to the standard of new software in many enterprises. The JavaBeans architecture truly allows for components based on components. This is not to say that all code must be written in Java language; the success is the platform-independent Java byte code binary. If these are not feasible, it is essential to at least use of the rest of the solution to the Java package separated by the native code. Data: XML and vocabulary XML obviously new appearance, it is a simplified SGML - 80% function and 20% complexity. The computer boundary has been widely used, and various signs indicate that it will dominate the data format in the future. Safety: The public key is not "clear" to send all key information, and the security system based on the public key has occupied dominance, especially in the Internet. Figure 3. Techniques that have been developed into standards from technology to users have changed the requirements for commercial solutions with the computational new world. In the past, it was built according to the client proposed to request the request. Focus on the focus in using the solution, and where they are, such as internal networks, external networks and the Internet. However, progress has indicated now focus more focuses on all data modeling and defining users and data. The computing solution has an inverted method, focus switches from technology and systems to information and users. Today, define a new solution involves defining the relationship between the user and the information subject. In most cases, a given information body has multiple users. Therefore, for an online store, when the customer views information, only give them access to certain specific data, and can be displayed in a suitable customer. When the supplier's customer service personnel view the same information, the viewing and display modes are different. From the perspective of solutions and users, this is a transformation, and it is the biggest challenge to today's IT industry. Users can continue to use it, because all techniques in the new tradition are indeed mature and proven, so the transition is one of the key points and strategies, rather than jumping to unknown technology. Parallel World can master these five basic technologies can bring another benefit. For many users, migrating to the new e-commerce world is progress, but not revolutionary. They can complete the most important transitions without having to give up all the investments they have already done. Therefore, this new world is actually a parallel world, not alternating the world. So, why will Java and XML succeed? There are several such reasons: It has been confirmed: all five parts of the new foundation are based on the most old and most confirmed ideas. TCP / IP, "Dumb" virtual machine, markup language, public key system - all decades of use, has been fully proof. User Drives: In the final analysis, the new basic movement is the market demand and expectations, not a supplier or a association recognized. Since the cost of computer technology is gradually focused, and the demand for new foundations is increasing due to the increasing pronnects caused by the update replacement. Supplier support: These five technologies have almost constitute the foundation of all suppliers' new solutions. Vendors trying to use alternative methods have gradually discovered that the market will ask questions about their choices, and they will suspect that they have a patented locked attempt. The platform has nothing to do: these five technologies are independent of each other and are independent of their platform. Thus, they can be performed anywhere as long as they will be separated from their systems and interoperability. Suppliers have nothing to do: These five technologies are not subject to any supplier control, not only from supplier locking risks, but also from any supplier of rewards that start upgrading competition. The only possible accident is Java and public key, which is worth spending time to consider in such an environment, why are these two are not a problem.
