Commentary on CODASYL Systems Committee's interim report on distributed database technology
暂无分享,去创建一个
The author has reviewed the Interim Report on Distributed Database Technology and is responding from the vantage point of the ANSIIX3/SPARC/Study Group on Distributed Systems. This study group is attempting to establish an architectural approach which will put all aspects of distributed systems into perspective, including distributed databases. The author has been studying this problem as part of Honeywell Information System's investigation for the last three years. My first comments must be addressed to the overall document which is well written and conveys its message well. I am in agreement with everything in it and will address myself to those aspects which it does not cover at the present time. As an interim report, one cannot criticize it severely for having some gaps. It has been circulated for comment and is soliciting comments so that the final report may be more accurate and more complete. I recommend the reading of this report to all interested parties and urge each reader to comment so that the final report will be as complete as time permits. The interim report lists a number of motivations for distributed systems. All are valid in one situation or another. There are a couple of additional reasons which I would like to add to the list. The first reason is a human factor reason. This is to permit some element of the business to physically possess the part of the corporate database which holds its part of the data. This gives them the responsibility and the satisfaction of updating their database without removing it logically from the required access by others. Another human factor reason relates to the distribution of the database to reduce its vulnerability to strike action or acts of terrorism. Another reason is the matter of database size vs. computer power. There is no computer on earth which can by itself handle all the data processing for a large corporation. Therefore, processing must be distributed and is already distributed for most of the FORTUNE 500 companies. In most cases they are already physically distributed but not logically integrated and will not be logically integratable until "distributed database facilities" become available.