From: "M. Edward (Ed) Borasky" <znmeb@...>
Date: 2007-07-08T10:35:24+09:00
Subject: Re: Oh the pain! Distributing Ruby apps

Sharon Phillips wrote:
> I work in a classified environment, so I can't give too much detail
> about what I do, so here's a vague outline:
> I work as part of a small team that supports a number of applications,
> most of which run on unix boxes. These applications are reasonably
> complex and busy.
> I'm not interested in CPU or mem usage (we have unicentre and other
> dedicated teams to monitor and manage these) but rather specifics within
> the system. Can't give examples. Sorry.
> There are times when things break and we have little warning, if any,
> but some type of dashboard would have given us the ability to preempt
> some of these. That's what I'm trying to build.
> Nothing can be added to the servers. I'm proposing something that runs
> on our machines.

Actually, that helps a lot. Since "nothing can be added to our servers",
that means you either have to instrument the clients or intercept and
interpret the client-server traffic, right? Since it's classified, I'll
go with "instrument the clients". :)

Given that, I think something like Distributed Ruby or Rinda might be a
workable base for this. Or maybe "event machine"?

 > Thanks to everyone else for the suggestions too.
> I think I'll have a closer look at JRuby. This seems the best fit
> currently, and I understand it can bridge to Java? That would be useful.
Yes, jRuby can bridge to Java.