Dr. Aniruddha Roy
ESRI India, NIIT GIS Ltd.
New Delhi 110019
Internet GIS software should support a range of configurations that support a diverse set of hardware and networking architectures.
Internet is unleashing profound changes in the society. Average selling price of PC has been sliding for 30 months in a row after 16 years’ stability. The top reasons for 85% of individual PC buyers presently are Internet. Applications like mail, searches etc. have more users than any Office suite of applications. This Internet revolution has resulted into the competition amongst the businesses to provide as much as information to the end users in a lucid and illustrative manner.
Map making and geographic analysis are not new, but a GIS performs these tasks faster and with more sophistication than traditional manual methods do. While the most common product of a GIS is often presented in map form, the real power of GIS lies in its ability to analyse.
Without owning the individual software, the Internet GIS technology has opened new paths for disseminating, sharing, displaying, and processing spatial information on the Internet. Web-based solutions provide a low-cost, efficient way to deliver map products to users.
The most common queries people would like to pose in the Internet using GIS functionality are for location display. Next popular event in the Internet is to show the driving directions along the streets. This routing information shows the shortest distance by time and distance impedances for a city and the information is updated regularly based on the traffic conditions like www.mapquest.com. There are sites like www.weather.com wherein one can see the low orbital satellite data showing weather condition and this gets updated every one hour. The users can see the dynamic changes of weather using the interactive player for last several hours. The latest of all the applications, in Internet is for the users to access GIS data at the browser end and integrate that with local specialised application data of their own residing in the local machine. This is now changing the entire model of working with GIS.
Internet Map servers
The first known geographic web site is that of Xerox (mapweb.parc.xerox.com/map) which was established in 1993. It is the world map that has been put in this site with options for many zoom levels. It is not with any true Internet map server but the site had pre-created a large number of different image maps for each case scenario So the request finds the particular html page containing the map and displays it back to the browser. However this was the good beginning and shown lights to several organisations to come forward to show their ability.
As the first to launch Geographic Information System (GIS) capabilities on the World Wide Web, using the middleware program, Internet Map Server (IMS), ESRI (Environmental Systems Research Institute Inc.) built its next generation Internet GIS based on years of expertise. Internet map server is also a very efficient way to make live GIS products available to a large population at minimum cost.
The initial release of the ArcView IMS ver. 1.0 from ESRI has been appropriately used by the Andhra Pradesh Roads and Building department, Government of Andhra Pradesh in 1997-98 for display of the road segments of the Andhra Pradesh with details of the type of the road, road condition, contractor for a certain stretch, engineer responsible for the task etc. That was indeed the first successful Indian Internet GIS project. The application was developed in ArcView GIS and served using extension product Internet Map Server using HTML.
To support the developers who have built their own custom built GIS application in MapObjects using Visual Basic, Visual C++, Delphi, Borland C++ Builder, Visual FoxPro, and Power Builder the Weblink facility was provided in the software MapObjects IMS 2.0.
ArcIMS 3.0 the third generation IMS product from ESRI has flavour of both desktop applications like ArcView GIS and developers choice of customising the same. It can serve in diverse applications as e-commerce, location based services, Enterprise Resource Planning (ERP) etc. The installation is simple wizard based with several standard templates for ready to use application. The MapMachine site is accessible to anyone with an Internet connection (www.nationalgeographic.com) and provides dynamic thematic maps, showing natural and human phenomena at a global scale using ArcIMS 3.0 internet GIS software. The GIS data is more than 150GB to support the generation of more than 500,000 customised, GIS-based maps every day.
System Design Architecture
To create an Internet GIS application and serve that on the net, primary factors like the type of GIS functionality, downloading time into the client browser and the cost of the whole application are considered. System design solutions must be based on user requirements. GIS environments require customer investment in people, data, applications, and hardware resources. It is important to make appropriate investments in each of these resources for GIS operations to be successful. The success of the GIS operations will be limited by the weakest component within the total solution. All the other software, hardware, and network components in the system must also be capable handling any expansion. The importance of selecting well-established software solutions and optimising all the system components cannot be overemphasised since all parts of the distributed configuration are critical in supporting the system’s overall performance.
After the GIS users, data requirements, and application strategy are identified, an appropriate software, hardware and network infrastructure can be defined to support GIS implementation. The general guidelines for the same is described below:
Internet GIS software should be hardware independent. As hardware manufacturers and technologies change over time and some manufacturers provide better functionality, performance, or communication capabilities as time evolves, the end user should look for the software that support multiple platforms. It should continue to support a wide variety of hardware platforms and to take full advantage of improvements in hardware, communications, networking, and database functionality, as they become available. The software should be in a position to support both Windows NT/Windows 2000 and also UNIX platforms.
The best system architecture design solution depends on the distribution of the user community and the type of operational data in use. A single map server can service hundreds of map products per hour to clients with standard Internet browser software. GIS products can be easily distributed to every desktop in an organisation (through the Web server), and to the general public through World Wide Web communications.
Hardware performance has improved rapidly over the past five years. Hardware pricing has dropped at roughly the same rate. It is important that organisations purchase what they need only when they need it, enabling them to benefit most from new technology. Computer hardware shelf life has reduced to roughly three years as per the recent surveys.
For Internet GIS application usage, one server class machine can be used for the functionality of the web server, map server and the data server, which can reside together. However, if the application is sophisticated and performance is the issue then web server, application server and data server can be separate. Application servers can be UNIX or Windows NT platforms that host application processes to a number of GIS clients. These platforms include terminal servers and Web transaction servers (i.e., map servers).
Fig. 1: ESRI Web-Server Solution
Web Transaction Processing
The Internet and Web technology have introduced a new processing environment to support GIS users. This environment allows a single application process to provide simultaneous support to a large number of concurrent GIS users (see Figure 1). ESRI’s IMS components include a map server installed on a Web server that directs inbound map requests to waiting application background processes for servicing. Each request is serviced and returned to the Internet browser client. Map servers can be scaled to support large volumes of user requirements and requests. This technology can very effectively put GIS map products on every desktop within an organisation using a single IMS server license and standard Web browser clients.
For several years, network technology was a relatively static environment while computer-processing power was increasing. Recent advances in communication technology support a dramatic shift in network solutions, introducing worldwide communications over the Internet and bringing information from millions of sources directly to the desktop in real time. Networks should be configured to accommodate GIS power users whose data transfer needs exceed typical GIS user bandwidth requirements. Access to common data sources for real-time display and analysis puts large demands on network communication, since data must move across the network to where the program is executed in order to display the information.
Performance on Internet
The performance of the system is measured by the ability to respond to the requests faster and the reliability of the system. Typically in the Indian scenario if the response to a certain query takes longer than 20-25 seconds the surfers loses patience and move on to other site. The overall system performance of Internet GIS is dependent on the combination of client, server, and networking performance – not the individual components. A combination of server-side and client-side processing that properly partitions the applications between the server and the client has a potential to have the best performance. GIS data uses both vector and raster data, which are large in volume, performance, becomes one of the most important issues in selecting an Internet GIS program. Preferably the applications should be tested using the Web Application Stress Test and data Analysis tools to know the response time for each individual module. This tool can be downloaded freely from the net from the site webtool.rte.microsoft.com.
Depending on the amount of processes performed on the client side, the client could be “thick” or “thin”. For the ASP (Active Server Page) kind of client it is Very Thin Client and the sizes are 50 – 75k with images. All processing is done on the server side. It supports very low bandwidth access and has fast downloading. It supports lower version browser. For HTML/DHTM clients it is Thin client configuration with size of 150 k. Server side execution happens here. It also supports low bandwidth access. It downloads quickly. For Java Viewer it is Thick client with size of around 750 k. Client & server side processing happens. Rendering, Buffering, Queries happens on the client side.
The network or communication performance in Internet GIS is affected by three major factors: network speed, networking software on both ends of the network, and the network traffic. A faster network clearly has better network performance. The speed of Internet connections is improving with faster modems and faster communication connections. On the other hand, efficient networking software on both ends of the network and an efficient design of Internet GIS programs can accommodate a slow speed connection to some extent. But the number of trips across the network is usually the most important factor in performance.
Reliability refers to the ability of the system to handle increased workload, such as multiple simultaneous requests. The normal assumption taken into consideration for the scalability factor is done by taking into consideration the Pareto’s rule which states that 80% of web traffic occurs during 20% of the time. To keep a system 99.9995% reliable one has to has redundant everything (hardware, software, network, etc.)
How many hours would it take for a hacker to hack through the firewall, is the normal question. The server should be protected using various mechanisms. The Internet GIS software should have the administrator password and user log in validation. The frequently method used for the protection of the map/web servers are by providing firewall. Firewall bandwidth is usually 80-90Mbps with single point of failure. It is simpler to put all machines either behind or outside the firewall. Security is less of an issue for server-side applications because there are no program codes that are executed in the user’s local machine. However, it is a concern for client-side applications, which are downloaded from the network, often from machines over which the user has no control. Such application codes could contain fatal flaws or viruses that could damage the user’s local machine.
Fig. 2: Gujarat Earthquake Rehabilitation Information System
Data sharing is important for any GIS program. Lack of data is the only hindrance in the awareness of this technology. The Geography Network (www.geographynetwork.com) is a global network of geographic information users and providers. It uses the infrastructure of the Internet to deliver geographic content to user browsers and desktops. Content may be provided in the form of raw data, maps, or more advanced services dealing with, for example, demographic mapping, flood risk mapping, and vehicle routing etc. The main theme of the Geography Network is the sharing of geographic information. ESRI maintains the Geography Network channel guide on a network server node.
The Geography Network is an open system. Anyone with an Internet browser can use the network, and becoming a provider is free and simple. An open protocol is used for communication that is compatible with emerging Internet standards for geographic information sharing. Indeed, the Geography Network technology has been used in the Open GIS Consortium Web mapping test bed.
The Geography Network is built on the technology of the Internet and the World Wide Web. Any user with a conventional Internet Web browser can access the Geography Network. Desktop GIS software such as ArcInfo, ArcView GIS, and ArcExplorer (freely downloadable from the site (www.esri.com) can also access data and services in the same way that they work with local data.
Out of the many applications that have been developed by ESRI India team for the web application, the most recent application for the UNDP is the application for the Gujarat Earthquake Rehabilitation Information System (GERIS) (Fig.2 and 3). The application is running in the web site www.undpquakerehab.com. The Bhuj earthquake that shook the Indian Province of Gujarat in India on the morning of January 26, 2001 (Republic Day of India) is the most deadly earthquake to strike India in its recorded history. One month after the earthquake, official Government of India figures place the death toll at about 20,000 and the number of injured at around 166,000. Indications are that 600,000 people were left homeless, with 348,000 houses destroyed and an additional 844,000 damaged. The Indian State Department estimates that the earthquake affected, directly or indirectly, 15.9 million people out of a total population of 37.8 million. More than 20,000 cattle are reported killed. Government estimates place direct economic losses at $1.3 billion. Other estimates indicate losses may be as high as $5 billion.
The server configuration is Pentium III with 256MB RAM. The application runs on Windows NT server 4.0 with IIS 4.0 and ArcIMS 3.0. The application has been developed using ASP and is very lightweight. The server is placed in the Gujarat Chief Minister’s office in Gandhinagar.
The main application of this GIS site is to provide the users with the map information along with the estimated loss of lives and property up to village level. RESECO, Gandhinagar and UNDP, Bhuj provided the GIS data. The application has the hierarchical database of District level, Taluka level and village level maps for the entire Gujarat, which can be accessed from the drop down menu.. The common GIS functionality like Zoom in Zoom out, Pan, identify, hyperlinks to the respective taluka level with the ground photographs of the disaster has been included. Other facilities like Print map and e-mail map is also provided.
Certain standard queries like, Number of Orphans, Total number of deaths, houses damaged etc. has been built as a standard dropdown query. Communication between the client machines and the server takes place using an open, published XML called AXL (Arc eXtended markup Language). AXL allows geographic information to be transferred over standard Web connections. It defines the request for content for a specific area and the resulting map or data content.
Fig. 3: Gujarat Earthquake Rehabilitation Information System
Successful implementation of a GIS depends on a good solid system design, appropriate hardware and software product selection, and successful systems integration. Internet GIS software should support a range of configurations that support a diverse set of hardware and networking architectures.