linphone-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Linphone-users] minimal flexisip proxy configuration?


From: Brian J. Murrell
Subject: Re: [Linphone-users] minimal flexisip proxy configuration?
Date: Sat, 01 Jun 2019 07:28:42 -0400
User-agent: Evolution 3.32.2 (3.32.2-1.fc30)

On Fri, 2019-05-31 at 08:04 -0400, Greg Troxel wrote:
> 
> Not answering your question, but could you either explain the theory
> of
> why one needs a proxy in front of a PBX,

Because currently most PBXes (Asterisk included) have no concept of
transient/mobile devices that will sleep and disconnect themselves from
the network and need a "push" to be woken up before they will be ready
to receive an INVITE.

And yes, I am aware of the IETF's work in this area but TTBOMK there
are no actual implementations on either the SIP server/proxy side or
the client side, so for practical purposes, it's still just "a paper".

While I have hacked up some pretty decent (IMHO) push handling for my
Asterisk instance, it needs a customized build[1] of linphone to work
reliably and even then there is a high latency (2-4 rings heard by the
caller) before my mobile phone even starts ringing.

Whereas my experience with sip.linphone.org, which uses flexisip, is
that it's much more responsive with much lower latencies between the
call being sent to it and the mobile phone ringing.  They have already
done all of the heavy lifting of figuring out the handling of mobile
phones with push, etc. in a reliable manner that I simply want to take
advantage of that and not re-invent that wheel (even though I have for
the most part, but that was more out of interest and learning than a
desire to maintain such a thing on my own).

> I am not entirely up to speed on SIP practices, but
> it
> seems like there is a culture of having lots of proxies far more than
> I
> would have thought necessary.

I don't think that's particularly true, particularly if you consider
that "SIP server" and "SIP proxy" are pretty much the same thing, as I
understand it.  So many people referring to SIP proxies are just
talking about PBXes through which they bridge SIP clients rather than
having the SIP clients talk to each other directly.

> A parallel puzzling web notion would be
> that you can't use a browser to look at a website without a local
> squid
> and a remove

Remote?

> nginx front end.

I'm not sure I see where the "remote nginx front end" is in the:

mobile_phone <-> flexisip <-> PBX

analogy.

But even otherwise, your analogy might be more accurate if you thought
about it as:

web_browser <-> HTTP2-to-HTTP1_proxy <-> HTTP1_website

where you stand up an HTTP2 proxy so that your browser enjoys the
benefits of version 2 of the HTTP protocol and lets the proxy handles
all of the "old crufty" version 1 HTTP interactions.

Not a very strong argument, given that all browsers that support HTTP 2
still handle version 1, but imagine a time in the future where a
browser drops support for version 1.  In any case it's just a more
parallel example of why one needs something like flexisip to handle the
complications of devices that most PBXes were just never designed to
handle.

Cheers,
b.

[1] A build of linphone that re-registers when it's woken by a push
message, even if it's already registered and within it's expiry, simply
to notify the PBX that it is awake and ready to receive the INVITE so
that the PBX doesn't send the INVITE too early and linphone misses it.

Attachment: signature.asc
Description: This is a digitally signed message part


reply via email to

[Prev in Thread] Current Thread [Next in Thread]