The hardest aspects? Authentication, (Who are you? Really.) Security (You can't tell what I'm doing or what data I'm working on, unless I let you.) Progressive abstraction and exposure (anyone should be able to understand how the system works and how to use it at an appropriate level. It's like physics. Your average person understands that when you throw a rock at something it will fly and possibly hit the target. A classical physicist understands the effect of the air, gravity, and initial velocity of the rock. A quantum physicist understands the chaotic nature of the rock at the atomic level and knows when to use simplifying assumptions (non-relativistic velocities, gravity, and non-quantum scale objects).) How do you expose the power of internet scale in ways that users of different levels can understand and use? Charles
Charles sscd (8/28/2009 9:19:49 AM)

First, designing and building robust services on 2,10 or 32 systems is easy. How do you build a system that can scale to thousands of systems? Second, how do you manage those thousands of machines? In building robust systems you have some interesting data management problems. Third, when you realize in thousands of machines and thousands of disk drives something will be dying all the time, how do you manage self-repair? Fourth, how do you find those subtle "bugs" that happen when you harness the power of all those machines? I can go on, but ... Chuck Wegrzyn
Chuck Wegrzyn sscd (8/11/2009 2:15:44 PM) Team Questions