OPTIMIZING CLIENT-SERVER COMMUNICATION FOR REMOTE SPATIAL DATABASE ACCESS
MetadataShow full item record
Technological advances in recent years have opened ways for easier creation of spatial data. Every day, vast amounts of data are collected by both governmental institutions (e.g., USGS, NASA) and commercial entities (e.g., IKONOS). This process is driven by increased popularity and affordability across the whole spectrum of collection methods, ranging from personal GPS units to satellite systems. Many collection methods such as satellite systems produce data in raster format. Often, such raster data is analyzed by the researchers directly, while at other times such data is used to produce the final dataset in vector format. With the rapidly increasing supply of data, more applications for this data are being developed that are of interest to a wider consumer base. The increasing popularity of spatial data viewers and query tools with end users introduces a requirement for methods to allow these basic users to access this data for viewing and querying instantly and without much effort. In our work, we focus on providing remote access to vector-based spatial data, rather than raster data. We explore new ways of allowing visualization of both spatial and non-spatial data stored in a central server database on a simple client connected to this server by possibly a slow and unreliable connection. We considered usage scenarios where transferring the whole database for processing on the client was not feasible. This is due to the large volume of data stored on the server as well as a lack of computing power on the client and a slow link between the two. We focus on finding an optimal way of distributing work between the server, clients, and possibly other entities introduced into the model for query evaluation and data management. We address issues of scalability for clients that have only limited access to system resources (e.g., a Java applet). Methods to allow these clients to provide an interactive user interface, even for databases of arbitrary size, are also examined.