1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
|
A simple distributed computing example with "pull" model.
There is a single central work dispatcher/gatherer that is contacted
by every worker you create. The worker asks the dispatcher for a chunk
of work data and returns the results when it is done.
The worker in this example finds the prime factorials for the numbers
that it gets as 'work' from the dispatcher, and returns the list of
factorials as 'result' to the dispatcher.
The client program generates a list of random numbers and sends
each number as a single work item to the dispatcher. It collects
the results and prints them to the screen once everything is complete.
*** Starting up ***
- We're using a Name Server, so start one.
- start the dispatcher (dispatcher.py)
- start one or more workers (worker.py). For best results, start one of
these on every machine/CPU in your network :-)
- finally, give the system a task to solve: start the client.py program.
Note: The dispatcher is pretty braindead. It only has a single work and
result queue. Running multiple clients will probably break the system.
Improvements are left as an exercise.
Note: because the workitem is a custom class that is passed over the network,
we use a custom class deserializer to be able to get it back from Pyro.
|
