The Sun Ray's big strength: server based everything, is also its biggest weakness. For example, home use of Sun Rays works beautifully for most business applications because the bandwidth and server computing loads for those are relatively minor. What happens, however, when the consumer wants to watch a movie is that both bandwidth and processor requirements peak - meaning that consumer bandwidth costs quickly become prohibitive.
The reason for this is that the Sun Ray really is just a remote console to processes happening on the server. This is great for high security applications and nicely facilitates Sun Ray's ability to preserve and resume sessions independently of what happens to the display, but makes it unsuitable for entertainment and related graphics intensive uses.
The same problem applies to desktop jobs like 3D automation and modelling: bandwidth and processing burdens increase dramatically because almost everything has to be done on the server.
The older smart displays, machines like NCD's X-terminals or Plan9's gnots, didn't have this weakness because display processing was done on the display, not the server.
Sun has two technologies that could enable a new generation of 3D, and consumer video, capable gear: NeWS and MAJC. The Network Environment Windowing system [NeWS] downloaded postscript to the terminal which then did the final processing - meaning that displays based on this approach could could use a local array processor (i.e. a GPU) to handle home video or 3D display for engineering software used in the office.
The MAJC CPU was intended specifically for desktop use with extensive GPU style scaleability and multi-media instructions.
I believe NeWS fell victim to Adobe's licensing demands on PostScript, and MAJC fell to a perceived lack of market demand for a Sun desktop.
However, there are three cost factors justifying a second look:
- Adobe is feeling increasing revenue presure from Microsoft and may now be amenable to a market expanding BSD style license on PostScript I and II;
- the MAJC CPU is essentially a free good for Sun developers because its costs have been fully written off; and,
- The SMP/CMT technologies now on the market as the T1 "Niagara" have demonstrated Sun's ability to offer dramatically improved user service at lower costs than ever before.
As a result benefits like the ability to provide 3D engineering and home video user support with relatively low bandwidth and server utilization via a low security Sun Ray variant may now be "low hanging fruit."
Since Sun hasn't yet announced anything of the kind, this section looks at the market and competition for such a product - including PCs, diskless workstation solutions like the LTSP, and specialized desktop appliances like those from Wyse and HP.
This leads to an interesting hypothesis about chickens and eggs:
- the thin client transformation is "in the bag" in the sense that doing it doesn't require much technical skill, imposes a minimum of organizational change, positions the organization for more positive change, increases both reliability and auditability, and saves money.
- the missing factor in pushing forward to widespread smart display adoption is leadership - and presure for change by user management. If, however, the thin client approach were widely adopted, we should see at least some organizations evolve toward the smart display approach.
In other words, if thin client gets you started, and some people will go the rest of the way by themselves, then the way to wide spread smart display adoption is through widespread thin client adoption.
- whether Sun does it or someone else does, development of a consumer oriented, but less secure, smart display for use by telcos and other home internet services suppliers seems inevitable.
- and once that happens, users who get thin client benefits at home, are going to drive experimentation at work.
[Editor's note: Although Paul Murphy is offline until August 7, he filed a series of chapter summaries for a book in progress on Sun Rays and the Smart Display Architecture.]