Dynamic Data Allocation in Model Computing

移动计算机的用户，如掌上电脑、笔记本电脑和个人通信系统，很快就可以通过无线网络在线访问大量数据库。这项活动的潜在市场估计每年可达十亿美元，包括接入和通讯费用。例如，在旅途中，乘客可以访问航空公司和其他航空公司的时刻表和天气信息。投资者可以访问金融工具的价格，销售人员可以访问库存数据，打电话的人可以访问交通信息，可穿戴的无处不在的计算机可以访问各种各样的数据库。无线通信比有线通信更昂贵，也更不可靠。我们证明了在移动计算机上分配/复制在线数据的方式会影响成本和数据可用性，因此数据分配必须根据不断变化的条件进行更改，例如访问模式、用户位置等。目前，分布式数据库的分配方案在设计时都是以静态的方式建立的。分配方案保持固定，直到设计人员手动干预更改副本的数量或位置。我们表明，这是一个不令人满意的解决方案，特别是在移动计算中。

Users of mobile computers, such as palmtops, notebook computers and personal communications systems, will soon have online access to a large number of databases via wireless networks. The potential market for this activity is estimated to be billion of dollars annually, in access and communication charge. For example, while on the road, passengers will access airline and other carriers schedules, and weather information. Investors will access prices of financial instruments, salespeople will access inventory data, callers will access traffic information, and wearable ubiquitous computers will access a wide variety of databases. Wireless communication is more expensive and less reliable that wire communication. We demonstrate that the way online data is allocated/replicated at mobile computers affect both, cost and data availability, and therefore data allocations has to change according to changing conditions, such as access patterns, users' location, etc. Presently, the allocation scheme of a distributed database is established in a static fashion, when the database is designed. The allocation scheme remains fixed until the designer manually intervenes to change the number of replicas or their location. We show that this is an unsatisfactory solution, particularly in mobile computing.