The computers that make it run

 
Bandwidth:
------------
 
low bandwidth:~5k/sec-user
users:40
Total bandwidth:200k/sec
 
More than the bandwidth of a business T1 connection. Assuming that we'll allow for expansion, high speed connections, and to not saturate the line (over 70%), we should go straight to a T3 to a NAP (Network Access Point).
 
 
High bandwidth:60k/user-sec
Total Bandwidth:1800k/sec
1.54 x 24 = 3600k/sec
 
Only 50% of capacity, which is acceptable. The web server and anything not required to control the bots in the project will be co-located at the ISP with the NAP connection.
 
Interfaces:
-------------
 
Low bandwidth connection is through a form with a cgi script that will control the robot. Star patterned buttons, and a visual update of the device location.
 
High bandwidth will be through a Shocked user page that will allow the user to control the robot using clicks, click duration, rollovers, and will be personally configurable.
 
Internal backend set-up:
--------------------------
 
Each server will be responsible for 5 robots, which will require 8 machines for robot control, a backup machine to quickly take over in case of hardware failure, a test server, and a dedicated logging machine. There will also be a dedicated machine to run the retrieval arm (which might be integrated with the logging machine), and a firewall.
 
This totals to 13 machines running Linux (a Unix OS for desktop PC architecture).
 
Each server will be responsible for testing the robot multiple times a day, keeping track of battery performance (so rechargeables get swapped out when they are starting to degrade), and general taking care the bots. They will call the retrieval machine when there is a problem, and basically mother their bots.
 
Site home