To: ftth who wrote (2007 ) 2/11/2001 4:17:04 AM From: Frank A. Coluccio Read Replies (1) | Respond to of 46821 re the Canarie wavelength drive, very interesting, but let's take a look. First, the intent is not to support desktop applications here, as much as I read collaborative computing. Clearly, the data sent over the wavelengths will not recirculate itself. I'm aware of optical circulators, and how they work, but those are different animals and don't apply here. So, the source must continuously transmit the same data over and over in a kind of broadcast and select scheme in this case. Whereby, all data is always being sent onto the lambda, whether anyone wants it or not. Users pick off only what they need. What's your take on this? Also, with multiple wavelengths at work in this fashion, the select process could be multi-tiered, tuning to lambda 1 for category of information 'a,' lambda 2 for 'b,' lambda 3 for 'c,' etc., and then drilling deeper within the desired lambda and selecting the specific data required, according to another addressing scheme at the file level. Even so, consider the refresh time (or the recirculating cycle). It would be long, even if the transmit speeds were 10- or 40- Gb/s [equal to 1.25- and 5.0- Gigabytes, respectively] if the data that was contained in a terabyte repository "back at the source" were being sent continuously. A terabyte being sent at 40 Gigabit s/s (5 Gigabytes /s) would take about four minutes to cycle, after overhead is factored in. A 500 GB source, 2 minutes. Wuddayathink? Can you deal with this as a replacement for your hard drive, or would it cause you to reach for a Valium? The way out of this is to use a much higher number of wavelengths, but then every PC would require an intelligent dwdm (an optical switch, effectively) attached to its bus. I can live with that as a prospect for the future. But today? Hmm...