Master: When the future is found online, it is obsessed that this article is found, it feels good, it is reproduced, from IBM. From the earliest use of file system storage data, the development of the database has been more than 50 years, and it has experienced the coexistence of the Hierarchical Database (IBM IMS) and a mesh database (GE IDS) in the 1960s. The Effect of Database of the Database of the Database, the Influence of Object Technology in the 1990s. 50 years later, the relational database is still in the mainstream position. After relationship technology, how should the database develop under the new situation? What is important trend? Recently, "Software Development" magazine has held an expert forum to discuss the future of the database. The forum is hosted by the consultant and writer, "Dr. Dobb's Journal" XML and Web Services Technology Channel Hosting, with Six Masters at the meeting: Rick Cattell, Don Chamberlin, Daniela Florescu, Jim Gray, and Jim Melton.
The discussion begins with the history of review. NORTH recalls that the price of an IBM 1410 computer has gathered from 300 wild horse sports cars, and now a double CPU with RAID server, the price is less than 10% of a wild horse motorcycle, everyone laughs. Relative to hardware development speed, the function of the database seems to be not fast enough. Undoubtedly, today, information storage, organization, management, and access is still one of the most important fields of computer technology. In recent years, with the rapid development of science and technology, e-commerce, in particular the rapid development of the Internet, the source, quantity, forms and methods of information have changed greatly, so that traditional relational databases face great challenges. The masters are looking forward to a variety of important trends for future.
Jim Gray believes: "Today, ordinary users also began to easily store and access any form of massive information on desktop computers, such as a song that you like, a video, a picture of a scan, These now don't save as files, you need to be able to store directly into the database. When searching, you can pass some identity text, or through the mapping, face recognition, or even OCR (optical character identification), etc. Natural ways. Microsoft currently has 3,000 development teams that are committed to putting all of these integration in desktop operating systems. In my group, three people have developed similar but relatively cheap versions. So, this dream may soon It will become a reality. Of course, the whole process is difficult. "
"This is indeed a big trend," Jim Melton agreed. "In fact, the structured information of the past database processing is different, we should first solve the data today, how to deal with complex type data: text, time, spatial, sound, graphics and video data, etc. The most effective way, Obviously, the database itself supports them, as a citizen in the built-in data type. "
Representatives from the object database technology camp do not leave the Bank: "Object database or object-relational database has been well implemented for these extended data types. Current databases have supported text and structured data The query, but there is also a gap between the probability reasoning of the information retrieval. "He also talked about the support for programming in the database. After all, he is the technical manner of Sun J2EE, which is the force behind JDBC and JDO. "Database has now joined the user-defined process, allowing to add data types with a certain behavior. However, the process and other executable code implementations such as triggers, are only two other citizens in the database, this needs to be done. Changes. "The Florescu from the BEA mentioned the currently hot sensor network plan, which will deploy the sensor of the huge nodes in a large geographic range to monitor the status and position of various objects in a certain area. This project itself contains huge potential. It is possible to open military intelligence and command, environmental monitoring, resource survey, meteorological forecast, traffic control, etc., and even change our daily life: if each object in the family There is a sensor, can't find or lose things, will become history. This is of course highly requested for database technology. Imagine thousands of, even more than 100,000, millions of sensors real-time flow information, what is the scene?
For P2P calculations, the client-server model has become a major data settlement program, and experts have achieved consistent. Chamberlin said: "P2P may be more suitable for updating less applications." JIM Gray believes that if there is no network bandwidth limit, P2P mode is possible because it requires a lot of data transfer. They gave negative evaluations for the so-called tether space calculation model.
In Jim Gray, lack of metadata is the main reason for current database technology unable to solve the problem in the Web. For example, find a photo or a piece of music that satisfies some conditions, it is difficult to solve it.
When someone asked someone asked "How to query in the document", Don Chamberlin did not allow the microphone. He compared structural data in traditional databases and non-structured or semi-structured, heterogeneous, and sparse data on the traditional database. He said: "Since the 1990s, it seems that all the computers are connected, and the information is unprecedented. However, these sudden information is not all consisting of rows and columns, with the order of order, There is also a streaming data such as a stock market. "There is a database query and document handling two aspects of the background of the chamberlin convinced that XML will be centrally located in the future database:" XML will become the main format of the document exchange. It is flexible And you can self-describe. Many programs need to be queried with the format of exchange data. Other data will be mapped to XML, linking with the outside world. But the relational database and SQL will not withdraw from the historical stage because they are independent of data. In terms of treatment relationships, there is natural advantage. "Of course, Chamberlin also admits:" There are still many jobs to do. For example, updates and transaction problems, XML data indexing problems need to be carefully resolved, this is also the W3C XQuery Working Group. Focus. We must build a sound infrastructure around XML as developing relational databases for many years. "
For the future development direction of the database, Cattel obviously has different views: "I don't think that the pure XML database will have a wonderful future. Do people use the XML format to store and query data? Performance is definitely big problem! I am 10 It was standing on the other side, and it was in favor of the problem-oriented technology to solve the problem. In fact, the current mainstream manufacturers have been well supported in the relational database. Interoperability and compatibility are of course important, XML as a general data The exchange model, because its scalability may be ideal for a variety of special purpose, but will never replace the relational database. "As an expert in Chamberlin intimate comrades and XML query engine, Daniella Florescu shows Myself for XML database confidence: "I think the optimization potential of XML queries is very large, and its performance is probably more than SQL. Currently, data groups are performed between XML elements, Java objects, and database table rows. It is the bottleneck of performance. This is the so-called 'web service impedance'. With my experience, I believe that if I use XML to program directly, I can overcome this problem. "Of course I can't take it in just 1 hour." The future development of the database is completely covered, but participants have benefited from it. Similar seminars will be held at the SD conference in March this year, let us pay attention.
Rick Cattell is an outstanding engineer of Sun. Hold JDBC API design, he founded an ODMG organization that had a great impact on database technology and served as the chairman, and is the world authority of the object database. At present, he serves as the main architect of Sun's Java Enterprise Technology Group, which is in charge of the future technical direction of J2EE that is involved in the creation.
Jim Melton has long been edited by ISO / IEC SQL standards, and the long-term members of the SQL Standards Committee have served DEC, SYBASE, and is currently senior engineers in Oracle. He is still aimed at combining SQL and XML project SQLX initiator.
Don chamberlin is a famous IBM, the father of SQL (see SQL's father: don chamberlin).
Jim Gray is the 1998 Tu Wen Prize winner. In the early years, IBM has participated in IMS, SYSTEM R and DB2 development, and has been served from Tandem and DEC. In 1993, Microsoft thought all the way, and even opened a research institute specialized in the Bay Area and finally invited him to join. SQL Server 7.0's success is inseparable from Gray's powerful support.
Daniela Florescu is a famous XML query engine developer XQRL CTO, and after the company was purchased by BEA, it entered the latter as a senior software engineer. She has developed XQuery language, member of the W3C XML Query Working Group, who has developed the W3C XML Query Working Group, developed XQuery language with CHAMBERLIN.
