Mosaic: Door to the On-line World

By Michael Harrington

Even when you set aside the media frenzy surrounding the so-called Information Superhighway, it is undeniable that the plain old Internet is gaining more popularity every week. As computers become the office tool par excellence for workers of all kinds, on-line access to information and communication with other people is a useful alternative to sorting through paper files and playing tag via voice mail.

But sorting through all this information has not been simple. Until now, searchers have had to access the Internet using different applications, such as FTP, Gopher, Telnet, news readers, and e-mail. The amount of time needed for a successful search can be substantial. A new tool was needed which would both combine existing applications into one tool and allow each organization to create a starting point for accessing pertinent Internet information. This tool is known as Mosaic .

It is both easy to use and inexpensive. In its most effective form, Mosaic is an easily readable page of text that contains highlighted words and images (see Figure 1). This format is familiar and approachable, resembling the printed article you are currently reading. Its second attraction is that, by clicking on these highlighted words and images, you can connect to other sources of information on your network or out on the Internet. These items can be other text files, images, movies, audio files, or other sources of information such as NFS file servers, World-Wide Web servers, Anonymous FTP sites, WAIS servers, Gopher servers, and even Usenet news servers. (For definitions of these resources, see the Glossary.)

By combining the familiar structure of a document with the capability to reach different types of data by a simple mouse click, Mosaic has established itself as the preferred Internet front end. This, of course, is beneficial to users who need better access to the Internet. However, its hidden attribute is that an organization can use this Internet tool to effectively access its own internal information resources.

When you click on the Mosaic icon, the application opens to what is called the home page. It resembles a GUI-based word processing document, perhaps including graphics and featuring formatted text, in particular underlined words. The word Mosaic is underlined in its first mention three paragraphs above. If you were reading this article via Mosaic (which you can; read on), you could simply click on the underlined word to get additional information from the creators of Mosaic. This uses a technology called hypertext, which gives you, the reader, access to additional sources of information without interrupting your train of thought. You have the option to go to the information if and when you want. Hypertext makes for an inherently robust and intuitive form of information.

To read this article via Mosaic, open the URL within Mosaic. You will then be connected to the UniForum World-Wide Web server, which contains information such as the 1994 Open Systems Products Directory, technical publications, and links to other sources of information.

If you do not have access to Mosaic, you can get the appropriate software from the NCSA's anonymous FTP site ( for your particular hardware platform. The operating systems currently supported are UNIX, Microsoft Windows, and Macintosh. As you continue to read this article, you will see other underlined words that are links to sources of information, which make understanding Mosaic easier. For those who wish to use other Internet tools, the URLs of these referenced sites appear at the end of the article.

The World of the Web

Mosaic can access many different types of servers on the Internet through uniform resource locators (URLs). This capability is what differentiates it from other Internet client software. URLs are flexible enough to allow Mosaic to access the many different types of servers available. Mosaic's development, however, is most closely related to the servers developed for the World-Wide Web project.

The World-Wide Web, or W3, is a true client/server application originally developed by the European Laboratory for Particle Physics (known by the acronym CERN) to meet the needs of the high-energy physics community. It has blossomed into a powerful global information system that makes various types of information on the Internet more readily available. W3 and Mosaic share roots in work done by Tim Berners-Lee at CERN, who wrote much of the client library and communication code that Mosaic uses. Although many different types of clients and servers fit under the W3 umbrella, the most flexible and popular pair is Mosaic clients and HyperText Transfer Protocol (HTTP) servers.

HTTP is the protocol that enables a Mosaic user to point and click on parts of a document to receive information from different sources. These servers require the involvement of technical personnel such as system administrators to install properly. But because Mosaic can be an effective client of many types of servers, effective business solutions can be implemented before having to deal with the issues of installing an HTTP server at your site.

Organizational Use

The National Center for Supercomputing Applications (NCSA) at the University of Illinois created Mosaic, as well as other useful applications such as NCSA Telnet. Since the NCSA is primarily funded by the National Science Foundation (NSF), most of these applications have been released into the public domain. Some of them are copyrighted but can be used freely by academic and research communities.

For-profit corporations may use Mosaic as long as they do so for internal business purposes only. The source code may be modified and distributed without cost as long as the NCSA is notified of the modifications and users of the software are clearly notified that they are using a modified version of the original NCSA distribution.

Mosaic and HTTP servers can more than adequately serve as data systems for the vast amount of information that many organizations maintain. Mosaic can immediately meet any organization's internal information needs. It will not replace mission-critical systems already in place, but it can act as a single point of reference for that information plus the large amount of "gray" information that accumulates within an organization. This information can be supplied from all parts of the organization including IS, sales, marketing, human resources, and operations departments.

Mosaic is best used to supply passive information, that is, information sought out by the users. This is usually information that does not change often, or if it does those changes are not critical. The user bears the responsibility to seek out this information. Schedules, meeting minutes, reference sources, policy handbooks, employee information, and benefit information are examples of passive information. (On the other hand, active information is given to users, usually whether they want it or not. A ringing telephone, e-mail, postal mail, and memos are examples of it.)

Cost Questions

The obvious appeal of Mosaic is that it is free and allows easy access to a vast amount of information already existing on the Internet. As with everything else, there is some cost associated with this freedom. But rather than being typical capital expenditures, they are employee-related. In fact, they are directly related to one of the downsides of the Internet, which is information overload.

These costs are incurred in three areas. First, the system administrator has to install and maintain the organization's home page(s) and/or internal server. Additionally, Mosaic must to be installed on the client machines. Although these processes are relatively straightforward, it adds yet another task to a probably overburdened employee.

Fortunately, a Mosaic-based solution can prove to be an effective method to ease some of that load. Putting almost all needed information in one location (such as a list of answers to frequently asked questions) can free the administrator from small interruptions. Because all systems function and get help in the same way, questions derived from platform differences are reduced.

Second, someone must manage the information for Mosaic clients. Although this person does not necessarily need technical skills, the responsibility often falls to the system administrator. It is important for information to be coordinated through one person in order to maintain a cohesive and effective resource. Since Mosaic is best at accessing passive information, this person should make Mosaic as appealing and well-known as possible. Structuring the home page to serve particular areas of business or research, types of information, or user types, and adding a searchable index are a couple of good ways to achieve this goal.

Appealing information is easy to find, accurate, and timely. Users must be aware of what is available via Mosaic, and its information must be immediately useful. If Mosaic becomes the single point of electronic reference for all types of information, it will infuriate users to page through a lot of irrelevant information to get to what they need%and then find that a hard copy hanging on a bulletin board is more up-to-date.

The Mosaic manager will need broad knowledge of the organization's business processes and information needs. His or her communications and listening skills will be important. Users will not use an information resource that is poorly organized, has outdated information, or is not properly advertised. In essence, this person will be the torchbearer of Mosaic.

Finally, there are technical issues of security, disk and CPU resources, and network bandwidth limitations associated with any client/server solution. To handle these questions requires a well-defined user base in order to predict peak loads and technical knowledge of the network topology and its current limitations. Therefore, it is best to create a solution for your internal network first and learn these lessons before you decide to offer information directly to the Internet. Once you have successfully completed this process, your team should have a much clearer understanding of the issues involved.

Getting Up

Assuming Mosaic is appealing as a tool, there are three steps to follow in implementing an internal Mosaic-based solution:

Mosaic is supported on a variety of UNIX hardware platforms as well as Microsoft Windows and the Apple Macintosh. There are some differences among the supported versions. NCSA supplies technical information on bug reports and work-arounds on each of the supported versions. They also make pre-release versions of their software available for those who are interested. System administrators should read the appropriate documentation that comes with each version to determine its functionality.

The easiest method to obtain the software is to use the FTP site mentioned before ( and go to the directory /Mosaic/Mosaic-binaries for the appropriate client binary. The source code can also be obtained at this site (/Mosaic/Mosaic-source), which you may modify according to the Mosaic copyright guidelines.

There are also unsupported versions of Mosaic, including those for Nextstep, DEC VMS, Novell UnixWare, Linux, HP-UX, SCO Open Desktop, AppleA/UX, and Amiga. Reportedly, their quality varies, and no help is forthcoming from NCSA.

As noted above, the first page of information a Mosaic user sees is the NCSA home page . You should change this default home page to one that you maintain at your site. Although the NCSA home page is adequate to use to connect to other Internet sites, your own home page will meet your organization's needs better and reduce the load on both the NCSA servers and the Internet. This occurs because every Mosaic client will be loading their initial page of information from a local file system or network server rather than connecting to the NCSA servers in Illinois. Figure 2 shows a simple example of a home page our laboratory uses and its necessary HTML commands.

There is a simple technical solution to implement a server with just ASCII files and the HyperText Markup Language (HTML). The caveat is that Mosaic clients need access to a common directory tree which contains your HTML files. As long as you maintain a global file system (such as NFS, NetWare, or AppleShare), this solution will work. Most networks are already set up in this manner. Once this basic solution is functioning, you can evaluate whether you should proceed to a more advanced solution, such as an HTTP server.

On the other hand, the internal server solution can be much more complex. For example, you may want to serve information to clients on different subnets or have separate servers for different departments as well as a general-use server for organization-wide information. For example, Russ Jones, Internet program manager for corporate communications at Digital Equipment Corp., currently supports 100 internal HTTP servers. These servers support information such as organizational charts, timetables for new products, engineering standards, gateways to legacy systems, and gateways to external stock quotes. He says that using Mosaic has greatly increased the productivity advantages for Digital's worldwide employee base. In fact, he is so convinced of this fact that he presented a case study on Digital's World-Wide Web server at the WWW94 meeting held in Geneva in May. (A second conference is scheduled for October in Chicago.) Digital's servers support 2,000 to 4,000 users per day. Digital has also announced that they and third-party ISV SpyGlass will bundle Mosaic with all DEC platforms to take advantage of Mosaic's power. The point is that Mosaic-based solutions can scale from simple LANs to enterprise-wide networks.

A Case in Point

At the Orthopaedic Biomechanics Laboratory at Beth Israel Hospital where I work, we began to use Mosaic in the fall of 1993 to solve the continuing problem of supplying solutions to the users of our Sun network. As the system administrator, I used to spend a lot of time pointing users to appropriate information about our network. Although this information already existed in different resources%including our laboratory handbook, a folder of important system-related e-mail messages, our annual report, Oracle database tables, and various items posted on bulletin boards around the laboratory%we needed a single point of reference to which a user could go to get it.

The Mosaic solution proved so effective that it quickly turned into our central repository for all information. For example, we provide PostScript files of our annual report, our laboratory handbook, and staff meeting minutes so they can be viewed directly on the desktop with a click of a button. We also have descriptions of our staff, with photos, to help orient the large number of students we support. In addition, all our laboratory's standard operating procedures for the established methods of performing research are available. Since our laboratory is located directly outside of Fenway Park, we even have a scanned image of the 1994 Red Sox baseball schedule to help us avoid traffic jams.

Besides the growing amount of internal information we maintain, we also have links to the rest of Beth Israel Hospital, Harvard University, and the Internet, including the National Institutes of Health (NIH) and their database of funding information. We also have links to our hardware and software suppliers to access marketing, sales, and technical information that they provide on their Internet servers. These include Advanced Visual Systems, Apple Computer, IBM, Mathworks, , WordPerfect Corp., and others. We find this a more effective resource than direct sales mainly because of its availability.

We now support an HTTP server that supplies our 30 Sun workstations with information. This server was originally installed on a low-end Sparc Classic but was moved to our Sun 4/470 file server when it became a production-level application. We have noticed little if any degradation of our network or the server(s) due to this application. The HTTP server software is available for platforms including UNIX, VMS, and Macintosh.

Currently our laboratory is leading an effort to establish a shared computer resource center among many research departments. Mosaic has become the tool to provide information to the many users in those departments. Our Research Computing Center home page now has links to each department's individual home page as well as a link to shared information such as software, shared resources, computer center information, and research-oriented resources. This method of supplying a single point of reference for many different departments creates an atmosphere of unity among them. Users continually respond with amazement as to the types of information we can supply via Mosaic. They're are also growing accustomed to my typical response, "Have you checked Mosaic for that information?" As a system administrator I am happy that I have to supply needed information only once, and users are happy because the information they desire is in a single resource that they are accustomed to using.

The capabilities of Mosaic to access all types of information and information servers are wide-ranging. It can display a movie as well as provide a gateway to an organization's legacy data. In addition, its point-and-click interface can be used by anyone who has experience using a graphical user interface.

Using Mosaic to access internal information not only streamlines the process into a single reference point but provides important information, such as hardware requirements, effective types of information to provide, and the nature of hidden costs needed to successfully establish an external Internet server. Again, a team with technical, business, and communications skills will be needed to enable your organization to define the infrastructure to support an external server while you wait for issues such as security, encryption, and public Internet access to be solved. For now, using Mosaic and WWW technology is reliable, the implementation is straightforward and scalable, and the results can be immediately rewarding.

SideBar 1

Ten Rules for Success of an Internal Server

1. Look at your department secretary's bulletin board for ideas on what to include on the server. Often there are one or two "go to" people who seem to know all that is worth knowing. Use them!

2. Keep the information up to date. If I log in three times and nothing is different, I won't be so willing to check again. One way to do this is to have your links be to other maintained sites so you don't have all the responsibility. Also you can maintain a "What's New" document which contains a listing by date of all items added to your server.

3. Maintain access logs to see what information users access.

4. Put some fun things on your server. In order to be successful, Mosaic must be used. Often at first that means just getting a user interested via information beyond work-related material.

5. Keep graphics to a minimum. Even though they look good the first time, the additional time to load the document becomes annoying not only to the user but to others on the network.

6. Keep marketing verbiage to a minimum. The best way to sell the information is to make the resource immediately useful.

7. Advertise an e-mail address directory for the person who maintains the server (Webmaster) and also for anyone the user may want to contact. If a user sees a problem with the server or thinks of something that would be useful, they will most likely send a message immediately rather than later.

8. Set up the address of your W3 server to be www.organization.domain so it can be easily remembered.

9. Include an index method to search your server for information.

10. Get a good draw/paint application and an image scanner. However, be aware there are copyright laws on using documents without permission.

SideBar 2


anonymous FTP
Process for document retrieval using the File Transfer Protocol; user logs on (as "anonymous") but does not need a password to access documents on the server.

Search device for documents on anonymous FTP server machines; the search matches file names. (Archie is to anonymous FTP servers as Veronica is to Gopher servers.)

Common extension for audio file names (

Another term for *viewer*; that is, a software package that permits you to look around the World-Wide Web. NCSA Mosaic is a browser.

Particle physics institute located in Geneva, Switzerland; home of the World-Wide Web.

Computer running client software that connects to server machines holding information; the client makes requests to a server for documents and is responsible for displaying the information. NCSA Mosaic is an example of client software.

File Transfer Protocol

CompuServe Graphics Interchange Format; common extension for a graphics file name (mygraphic.gif)

Menu-driven system on a server machine that presents files available for retrieval in hierarchical fashion; Gopher is considered particularly easy to use because of its menu system.

Term for the interconnected Gopher servers.

Home Page
Start-up document that serves as home base for Internet explorations.

HyperText Markup Language; the collection of styles used to define the various components (e.g., bulleted list, citation, emphasis) of a hypertext document.

Hypertext Transport Protocol; protocol used by World-Wide Web servers.

Word or graphic in a file displayed on screen with some form of highlighting (color or underlining or both). The word or graphic represents hidden text containing the URL information of another document, which is displayed when you click on the highlighted word or graphic.

Richly formatted documents containing hyperlinks.

Interconnected network of networks; provides easy, reliable connections to computers virtually anywhere in the world.

National Center for Supercomputing Applications at the University of Illinois, Urbana-Champaign; home of NCSA Mosaic.

NCSA Mosaic
Distributed hypermedia system developed by NCSA Software Development Group; designed for information discovery and retrieval using the global Internet.

Computer that provides services (documents, software) to other machines that run specific software (clients) and request those services.

Standard Generalized Markup Language; language that defines the structure of a document and the logical relationship of its parts to a source file that is formatted for display in your Document View.

Uniform Resource Locator; server and path information used in an HTML-coded source file to locate another document. URL is inserted in a document in the following format: scheme://host- domain[:port]/path/filename.

Wide-Area Information Server; index and retrieval system. When you enter a keyword, a search is performed on indexed documents, which then can be retrieved.

World-Wide Web; distributed hypertext-based information system developed at CERN.