VAN is all about running applications on remote servers and transporting the resulting stream of images to one or more client devices. The clients then decompress the image stream (if necessary) and display the interactively generated results while at the same time sharing keyboard and mouse control of the application with the visual server. All communication takes place using commodity TCP/IP networks at rates that need not exceed those of 100Base-T, ATM, or Gigabit Ethernet networks. In fact, even T1 wide area networks are quite capable of delivering interactive visual serving sessions across intercontinental distances. In this way, the desktop user can have access to advanced rendering techniques (such as volume visualization), integrated high-performance computing, hundreds of gigabytes of memory, and multiple gigabytes per second of I/O that are available only on Silicon Graphics Prism and SGI Onyx family systems.

• A visual server can be either a multi-user SGI scalable visualization system (like the Silicon Graphics Prism or Silicon Graphics Onyx4) or a Silicon Graphics visual workstation like the Silicon Graphics Prism Deskside or Silicon Graphics® Tezro® that acts as a personal collaboration system.
• A client system, which can be an SGI IRIX, Mac OS X, Sun™ Solaris™, Linux®, Microsoft® Windows NT®, Windows® 2000, or Windows® XP workstation or a wireless tablet computer running Linux or Windows® XP Tablet PC Edition
• A commodity TCP/IP network to connect the visual server to the desktop client
• OpenGL Vizserver software that runs on both the visual server and the client systems and provides compression, communications, and control over the network
• The user's OpenGL® applications, running without modification
• SGI Professional Services to customize the installation and integrate it into your operational environment

Individuals also benefit by having direct access to experts to help solve difficult problems. By using desktop-to-desktop personal collaboration, an individual can work with an expert in a specific field for a short amount of time without having to copy data or worry about software revision levels. By allowing remote experts to directly manipulate the applications and data on their systems, individuals can drastically reduce the amount of time required to solve specific problems. The cumulative effect of more efficient use of experts is that an organization is more productive and delivers designs and solutions in less time and with less cost than ever before.

• Their most important and expensive resources--their people--are more productive and generate more discoveries in less time
• A single, scalable system is deployed as a shared resource instead of requiring multiple dedicated SGI Onyx family systems
• The virtualization of visualization resources enables organizations to reallocate resources without moving them, thus maximizing utilization while meeting peak demands from individual groups.
• The same Silicon Graphics scalable visualization system can be used to drive an SGI Reality Center facility with all of its associated benefits
• The same system can be used as a supercomputer to solve large-scale problems, producing breakthrough results for the organization

• Videoconferencing: Videoconferencing uses low-resolution NTSC or PAL inputs to deliver video to a remote system. This input is, at best, 640x480 resolution and is encoded in a very lossy fashion, so it delivers relatively poor image quality. Because the content stream is unidirectional, there is no way to control the server from the client.
• Redirection of X and OpenGL displays to a remote system: This solution sends all of the graphics commands to the desktop system for local execution-thereby limiting image quality and graphics performance to the capabilities of the desktop system-while loading the network with every piece of information necessary to do the rendering. If the scene is simple and the network fast and otherwise inactive, this solution can work very well. However, today's leading applications use large amounts of texture memory, large numbers of polygons, and advanced rendering techniques that are unavailable on desktop systems. For leading applications and current problems, compressed images created by OpenGL Vizserver can now take up less network bandwidth than the underlying streams of OpenGL data used to generate them.
• Fiber-optic cabling to individual desktops: This solution provides high-quality images to directly connected desktops. However, this solution requires that special cabling be pulled to every office where the use of the scalable system will be needed. This cabling is limited to short distances of ~1 km. Visual serving solutions use general-purpose networks that are most likely already in place, thus eliminating the cost of pulling specialized cable and overcoming its distance limitations.
• Shared desktop capabilities (e.g., Microsoft® NetMeeting® or SGImeeting™): These solutions work in a manner similar to OpenGL Vizserver but are designed to work over low-speed lines, provide roughly 10% of the performance of OpenGL Vizserver, and do so with reduced image quality.

• Scalable computing, visualization, memory, and I/O: You can access large amounts of data at up to 300 times the rate available on desktop workstations or from a local NFS server (OpenGL Vizserver over a slower wide area network can provide you with even greater data access benefits), and you can extract analysis information from complex data sets at up to 50 times the rate possible with a desktop system
• Resource sharing: Many users need access to high-end applications or databases for only a fraction of each day; OpenGL Vizserver can substantially reduce the cost of application software, hardware infrastructure, and system maintenance while increasing access to these tools throughout the organization
• Single data set and protection of intellectual property: Organizations may not want multiple copies of their graphical data sets floating around, and they may not want graphics files being FTP'd, even within an organization, because of security issues

Additionally, OpenGL Vizserver provides a compression API. Using this API, a user/developer can integrate his or her own compression algorithm into OpenGL Vizserver. These algorithms can be based on frame-by-frame compression methods like CCC and ICC, or they can use interframe compression techniques like MPEG.