[Spread-users] Spread and queuing

Colin Meyer cmeyer at helvella.org
Fri Nov 4 17:04:32 EST 2005


I am new to Spread, and am exploring it for use in an application.

The problem I am trying to solve is communicating between a web
application front end (mod_perl) and many "data provider" backends.

The application accepts search requests and applies them against a
number of data provider backends, which may be local databases, other
local applications, or remote applications (REST and other). The data
providers each have a Perl based front end, which interfaces between
whatever the real data source is, and the API that the mod_perl web front
end knows how to talk to. Each data provider type will have multiple
child processes (so that many queries may be processed simultaneously).

I would like to define a Spread group for each data provider type, that
all of its children would join, and listen for requests. However, it
seems that Spread's behavior is to always send each message to all
listeners joined to the group. I would prefer to have a first-comer
gets to handle the request, while the other children keep listening for
other requests.

Can I have Spread act in this fashion, or am I barking up the wrong

And a side note: is it possible to increase the maximum message size
to larger than 100k by recompiling Spread, or should I just chunk 
large messages at the application level?

Thanks for your help,

More information about the Spread-users mailing list