Server types:
--------------------------
 (1) Iterative server: each client request is
      processed in its entirety before accepting
      the next client.  Use this approach if
      connection/service time is very short
      (avoids overhead of forking, etc.);
      also supports mutual exclusion for clients

 (2) Concurrent server: client handled separately
      from the listener; parent listens again
      immediately.

 (2a) Concurrent "fork" server: more overhead;
     if a child (parent) process dies, others are
     okay; may take advantage of multiple CPUs

 (2b) Concurrent "thread" server: faster than fork;
     but if the process dies, ALL threads are dead.

 (2c) Concurrent "select" server: less OS overhead
     related to fork/thread maintenance; used to
     listen on many ports

 (3) Pre-forking server or pre-threading server:
    create a bunch of child processes or threads
    and have this "pool" available when clients
    connect

 ---- could be static number of threads/processes
 ---- or could dynamically change based on traffic

---------------------------

 Design client/server diagram/architecture for:
 (a) a Web server with a database back-end
 (b) a distributed file system

 What kind of server?  Why?

 What "pieces" go where?  Why?

 Security? Performance? Extensibility? Reliability?



 Problems:
--------------
--> single point of failure
--> bottlenecks
--> server overloads
--> cost $$$
--> complexity

 Advantages:
---------------
--> extensibility
--> reliability
--> ability to swap in different components
        (e.g. database)