[openss7] LPCP's place in the world

[Greg Herlein <gherlein@herlein.com>]

The single most common question I get asked is "why LPCP?  why
not just use MGCP or MEGACO?"

It's simple:  LPCP is not a replacement for those protocols.  It
fits a particular niche in the bigger picture, with some, but not
much, overlap with other protocols.  Here is an FAQ style
discussion of LPCP and the LID/LCS architecture.  I hope you find
it interesting.

-------

Q:  Where does LPCP fit into a bigger world of established
telephony protocols?

A:  LPCP is a low level protocol.  In the MGCP/MEGACO world that
is most similar to the control model that LPCP subscribes to,
LPCP would be used to create distributed decomposed gateways.  
The Media Gateway Controller (MGC) would still be in charge, and
would control Media Gateways (MGs).  However, as noted in the
IETF draft to support IP Phones, there is a need to support
gateway systems that are "decomposed."  A traditional gateway may
very well be a monolithic system with line cards and the MG
software co-existing in the same box.  The IP Phone decomposition
describes a system where there is no real MG, but each IP Phone
would be a capable miniature GW itself.

Q:  LPCP seems a lot like the MEGACO IP Phone package.  Is it?

A:  LPCP, and the LID/LCS (Line Interface Daemon/Line Control
Software) architecture provide a slight variation on the IP Phone
model:  that of a decomosed gateway where the actual line
endpoints are nowhere near as capable as an IP Phone, yet still
capable of being collectively treated as gateway endpoints.

The model would look like this:

 Controller    Media Gateway         Endpoints 
   MGC <-----> [MG | LCS]  <-------> multiple LIDs 
	 
Now, the LID/LCS architecture *may not* even plug into any other
protocol suite - LPCP is flexible and powerful enough that small
IVR, PBX and residential phone systems could be built using just
LPCP.  And, due to the radically simpler design of LPCP over
MGCP or MEGACO, these systems can be developed, produced and
deployed faster and CHEAPER than if you used MGCP/MEGACO and
tried to deal with interoperability issues and full
compliance.  Those two ideals are certainly important, but as
shown in the diagram above, they can be encapsulated into the
MG/LCS component and isolated into one unit of software that can
easily be tested independant of the variety of endpoint systems
that are possible with LPCP.

Q:  Is LPCP in competition with other telephony protocols?

A:  NO!  LPCP is not in competition with other protocols.  It's
not out to be a replacement for MGCP or MEGACO.  It's existence
does not imply derision for those protocols, or for SIP or H.323
(though my views about H.323 are well known independent of my
LPCP work!).  


Q:  What is the reason for yet another protocol?  Why?

A:  LPCP is intended to be a simple way to build smaller-scale
phone systems that can be "plugged into" bigger phone systems
using those protocols that do scale to those sizes.  It's meant
to be EASY to code, CHEAP to implement on very low end hardware,
and easily EMBEDDABLE in that low end cheap hardare.

---------

I hope this will answer some folks questions about where LPCP
fits into the big picture of things.  I'm re-doing the OpenSwitch
web pages to reflect this more clearly, and I will re-state it on
the web site in a new FAQ I am writing.  

Look for a new LPCP document in the next few days as well.

Now, back to coding!  Thanks for reading!

Greg

/**************************************************************
Greg Herlein                                   greg@herlein.com
Herlein Engineering			        www.herlein.com
***************************************************************/