[storm] question about strategy

shawn bright nephish at gmail.com
Tue Jul 17 16:45:30 BST 2007


Hello folks!

i have a question about strategy. Our company takes data in from field
sensors several different ways ( serial port, ethernet, etc..)
and we process this information and send pages to customers when the status
of their machines change. We also use the info to show their stuff on a
website.

The whole this is run from a GTK2 app written in python on Ubuntu Linux.

To process everything coming in and all control going out, we are running
about 16 threads, all initiated from the GUI.

I am wondering what the best strategy would be for creating store objects.
Each thread runs in it's own class

like
class SerialDataIn(threading.thread):
   def run(self):
         while 1:
              process stuff comming in over and over forever, or until we
restart the program, or a hard drive melts

each thread will require at least a half different storm classes like
Sensor, Customer, etc.. and will process several hundred records each time
through the loop.

i know thats a lot, but i say that to ask this:
what would be the best strategy for where to put my store?
should i have several ? Like one per thread ?
or should there be only one that i pass to each thread from the GUI as
required?

i am looking at porting all of my app to storm because the way i am doing
things with SQL is painful. And i can do this one thread at  a time,
whenever i like.

Any tips?

thanks

shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.ubuntu.com/archives/storm/attachments/20070717/074f3ef7/attachment.htm 


More information about the storm mailing list