[Chicago] Erlang Envy, was Re: Stackless

[Kumar McMillan]

On 6/5/07, Pete <pfein at pobox.com> wrote:
> On Tuesday June 5 2007 5:03 pm, Kumar McMillan wrote:
> > On 6/5/07, Pete <pfein at pobox.com> wrote:
> > A huge challenge in decoupled SOA environments is sending a message
> > somewhere with a 100% guarantee that it doesn't get lost (when taking
> > rapture into consideration, 99.999% is usually acceptable).  This is a
>
> A good quote on reliability (2nd)
> http://patricklogan.blogspot.com/2007/04/systems.html

I find this stuff fascinating.  The pybots project [1] works like
this, like many cooperating cells (build slaves that pop on and off
here.  If a slave goes offline then another one picks up).  It's
possible to apply this concept to many things.

I read a great article [2] the other day that has a good quote about
distributed systems vs. SOA:

"...the move from distributed systems (one transactional scope --> one
notion of time) to SOA (independent transactional scopes --> time
based on the perspective of the user) is like moving from Newton's
Universe to Einstein's Universe."

specifically, when you have the concept of "one system" (it might
actually be multiple systems) then it's like operating in Newton's
universe where time marches on at the same pace for everyone.  But
then if you design the system to be made of many systems that do
useful things without worrying about each other it's like looking into
the night sky and seeing the distant light from a galaxy that might be
dead (Einstein's universe) because their "time" is not your "time."
(paraphrased from the article.)

[1] http://www.pybots.org/
[2] http://blogs.msdn.com/pathelland/archive/2007/05/20/soa-and-newton-s-universe.aspx