Tiger Video Server
------------------

1) Discuss the file layout for single bitrate files that ensures that failures
   of cubs can be tolerated without overloading the throughput capacity
   of each disk.

2) What are "hotspots" in the schedule?  How does the Tiger system overcome
   hotspots?

3) Explain the difference between single bitrate and multiple bitrate files.
   How does that affect the design of the disk schedule?

4) Given that there is no global state of the schedule, how does the
   Tiger system keep the cubs coordinated to deliver the multimedia 
   streams to the clients?  Explain the pros and cons of the implementation
   choices considered and the one adopted in Tiger.

5) Conventional file systems such as NFS keeps the metadata (file index
   blocks) on the disk.  How does the Tiger system from this conventional
   set up and why?

6) Write in algorithmic form the protocol for adding and deleting new
   viewers to the schedule in Tiger for single bitrate files.

7) Discuss the complexities in dealing with multi bitrate files in the
   Tiger schedule.


Rialto Paper
------------

1) Succinctly describe the abstractions provided by the Rialto OS.
   Explain why these abstractions are useful in the context of 
   serving multimedia applications.

2) What are the goals of the Rialto OS and how do they differ from previous
   works on supporting real-time constraints in OS?

3) Consider a scheduling graph with a base period of 10 ms as shown
   in Fig 3-1.  Two new requests come in:
      G: 1ms every 40 ms
      H: 1ms every 40 ms
   Show the new schedule.

4) Consider a scheduling graph with a base period of 10 ms as shown
   in Fig 3-1.  Assume the original request by A was:
      A: 6ms every 30 ms
   Does the schedule shown in Fig 3-1 really satisfy A's request?
   Explain using a timeline of the schedule why or why not.  
   
5) The scheduler is currently serving the following two requests:
      A: 6 ms every 10 ms
      B: 1 ms every 10 ms
    

   A new request comes in:
     C: 1ms every 5 ms

   Show how the scheduling graph is morphed to handle the new request.

6) What is the idea of the second chance queue in the scheduler
   implementation?

7) How are interrupts handled in the presence of the scheduling guarantees
   in Rialto?

8) Develop the data structures for the Rialto scheduler.