|
|
|
Invited Presentations
|
Two technical factors to migrate GIS into web environment
Shuichi Takino
Dawn Corporation, GIS Research and Development Dept.
Kobe KIMEC Center Building 4F, 1-5-2 Minatojima-minamimachi
Chuo-ku, Kobe 6500047
takino@dawn-corp.co.jp
As we enter the populalization period of GIS in Japan, the execution environment is migrating to
Intranet/Internet environment (Web environment). Basically, GIS manipulates vast amount of data and requires
a number of people to operate the system concurrently and cooperatively. As a result, GIS has developed as a
network-based computer system. Along with the network system transition from LAN environment to Web
environment, GIS is expected to migrate into Web environment. However, because of the Web-based system
restriction, conventional system is not able to realize preferable response time. We have invented a new Webbased
system and realized a drastic improvement.
This paper clarifies major problems of the conventional Web GIS first and later illustrates how to improve
them and the test results.
The followings are the problems of conventional Web GIS we have tackled.
- The limitation of GIS server transaction capacity for servicing concurrent multiple clients.
Generally, 10 to 20 clients are upper limits to be served by one GIS server concurrently.
- The response time problem of Web GIS operation. When you display maps or execute spatial query,
simple operation such as zooming in/out and scrolling binds you to wait for a next map to be desplayed.
- The difficulty in implementing spatial analysis function which is the basic element of GIS. Web GIS is
resigned only to map display and attribute information retrieval.
- The difficulty in implementing spatial data update function from Web clients, because simple exclusive
control mechanism is not able to meet market requirements.
Chapter 2 will discuss how to solve problem 1), 2) and 3). Chapter 3 will discuss how to solve problem 4).
Distributed System for Manipulating Geographical Data
Conventional Server/Client Model for Geographical Data Reference

Figure 1. Conventional Server/Client System for Geographical Data Reference
The upper figure illustrates the mainstream model of conventional Web GIS for geographical data reference.
Basically, this model uses server/client structure on LAN environment and adopts http protocol for
communication system. First of all, PC clients require specific kinds of maps and area to be displayed via
WWW server to GIS server. In response, GIS server loads geographical data files and convert data format so
that PC clients should display them. One data format is image data, and the other format is numerical data
which PC clients plug-ins are able to translate. Although the later model decreases GIS server process, both
model requires GIS servers to process geographical data one by one, and send processd data back to PC clients.
Generally, GIS manipulate vast amount of data and each PC client requires different area s of data. As a
result, GIS servers are overloaded for retrieving and converting large geographical data. This model is difficult
for processing and providing geographical data concurrently for multiple users thus leading to long waiting
time as illustrated in figure 2. In order to improve transaction performance, the only solution is the hardware
reinforcement such as installing another server machine.
Figure 2. Sequential Data Distribution by Serber/Client Model
|
|
|
|