[OpenSIPS-Users] About OpenSIP "The New Design"

[Dan Pascu]

On Thursday 27 November 2008, Iñaki Baz Castillo wrote:
> El Jueves, 27 de Noviembre de 2008, Dan Pascu escribió:
> > > Well, while it seems good, I consider it impossible
> > > (unfortunatelly).
> >
> > Fortunately, history shows us that progress was done by people who
> > didn't knew that something was impossible, so they tried anyway :P
>
> Sure something can be improved, but what I meant is that it's not so
> easy to separates layers when dealing with NAT issues.

I'd say let's leave this debate for when we have a design blueprint in 
front of us. Otherwise we can speculate endlessly.

> > > How would be the performance if OpenSIPS must run PHP/Python/Ruby
> > > code for each message?
> >
> > I'm sick and tired of people being performance experts without
> > running a single benchmark, just based on assumptions and common
> > myths. Not to mention that in this case it's even more silly as there
> > was no design presented yet, just some requirements. So you have no
> > idea how things will be, but you can already make absolute claims
> > about the performance of the system... I rest my case.
>
> Dan, if you re-read my phrase you will realize that *it was just a
> question*, I'm not doing assumpions.

I know it was a question, but my frustration related to the performance 
obsession on *SER lists has kicked in. Unfortunately performance as a 
reason was abused many times on the *SER mailing lists and has even led 
to bad design decisions. As I said many times, performance cannot be 
discussed without benchmarks and code profiling. And there are other 
things to consider as well along performance.

There's a story that reflects what I think about how the performance 
argument was used many times in the *SER communities in the past:

An absent-minded professor was getting late for a presentation he had to 
make. Realizing how late it was, he jumped into a taxi and yelled to the 
driver: "Hurry up! Hurry up! Go as fast as your can!". While the cab was 
running like crazy through the city, the professor realized he didn't 
mention where he wanted to go. So he asked the driver: "Do you know where 
you're going?". "No sir" said the driver, "but I go as fast as I can!"

> > > Also there are cool functions in OpenSIPS modules, for
> > > example "loose_route()". Would you imagine implementing that
> > > funcion in each "possible" high level language?
> >
> > Where did you read that loose routing will be performed by the
> > application layer?
> > And BTW, loose_route is not a cool function. It's an oxymoron.
> > Something that should be done automatically and you should never even
> > need to know about.
>
> Imagine you run the "presence" module in the same box as the proxy.

I can imagine many things ;), but as I said let's discuss on the design 
proposal when it becomes available.

> Then you do need some logic in your in-dialog section (but not into
> "loose_route" section since in-dialog SUBSCRIBE/PUBLISH will have no
> Route header pointing to our proxy). I think this is a mix between
> application/routing/core layer, is not?

All I can tell you is that the new design we have in mind is so different 
from what you already know, that it's very difficult to make assumptions 
about it from what you already know from the old design.

-- 
Dan