[OpenSIPS-Users] OpenSip SIP, SIP-I e SIP-T

[Alex Balashov]

To expand on this just a little bit:

While here in the VoIP cottage industry we mostly deal with SIP to begin 
with, in that we use ISDN gateways for connecting to carriers, get SIP 
trunking from our carriers/ITSPs, and so on, the reality is that most 
stuff in the PSTN carrier space is still done with big-iron TDM 
equipment, at least here in the US.  If you want to be a competitive 
carrier, you *must* interconnect with the incumbent telco using SS7;  no 
ands, buts, ors.

That doesn't mean there aren't a lot of opportunities to deploy SIP 
internally inside the service delivery core.  The main benefit SIP 
provides there is that it is so high-level and easy to manipulate.  As a 
result, a lot of mediation, logging, billing, analysis, translation, LCR 
  can be done easily and inexpensively.  Before SIP and H.323 came 
along, doing this kind of stuff required a box that did all that and 
spoke SS7 or, at the very least ISDN Q.931, and that is much more 
expensive, inflexible, and difficult to manipulate.

Promoting this traffic to a higher-level protocol stack that has more 
applications and tools to deal with it allows the development of 
solutions for doing sophisticated telco-world stuff using commodity 
hardware and open methodologies, open-source style.  That has triggered 
a wave of new products and paradigms in the telco space in a way that is 
analogous to how Asterisk et al have revolutionised the PBX space.

One example of this is TransNexus' NexOSS/NexSRS product 
(www.transnexus.com).  They use the OSP (Open Settlement Protocol) 
module for OpenSER and/or for Asterisk (depending on whether a B2BUA is 
required) internally inside their product to perform a lot of neat AAA 
and routing functions (e.g. the NexSRS route server).  Their ability to 
do this benefits precisely from the fact that the traffic can be moved 
onto a higher-level protocol plane and away from proprietary, expensive, 
closed and inflexible stuff that has been a defining feature of the 
telco world.  If you can turn the traffic into SIP or H.323, they can 
deal with it, but if it's SS7 or PRI, they can't.  The world is going 
more "soft[ware]."

At the same time, the telco space is not a SIP world right now;  the 
network edges are still SS7, and the market really hasn't settled on a 
good private SIP interconnection/peering strategy and implementation for 
intercarrier settlement. So, for the most part SIP trunking is used for 
customer access only.  The SS7 information must be conserved in this 
type of setup, and that's one of the reasons the sort of thing that 
SIP-T is exists.

Alex Balashov wrote:

> Adrian Georgescu wrote:
> 
>> Why should SIP-T still exist? Is it cheaper than having a gateway? What 
>> is the practical use case for investing in such technology?
>>
>> I am eager to learn
> 
> We've used it extensively in work with CLECs that operate TDM switches 
> such as the Metaswitch, Lucent LCS/Telica, etc.
> 
> When a carrier operates more than one switch, SS7 interconnection 
> between them is generally required so, for the same basic reasons an 
> internal iBGP mesh or partial mesh (confederation) between two border 
> routers is required for IP.   One switch must be aware of numbers routed 
> or ported into the other switch, and so on.
> 
> The reason for its existence is that if both network elements support 
> SIP-T, it allows you to replace an SS7 IMT (inter-machine trunk) with an 
> IP-based mechanism for this interconnection.  This allows you to move 
> the traffic over a data network and get all the benefits that this 
> brings;  economies of scale through decreased facilities, 
> oversubscription, etc.  The main benefit is the elimination of TDM trunk 
> exhaust;  SS7 IMTs are physically bundles (trunk groups/TCICs) of DS0s, 
> usually consisting of one or more T1s, and sometimes DS3s or more.  That 
> means that when a large volume of calls is running between the two 
> switches, you could burn up all your SS7 trunks.  Running the calls as 
> SIP-T allows you to use something like a gigabit network core to make 
> that problem go away somewhat -- a key benefit of VoIP in most other 
> scenarios with which you are familiar with.
> 
> At the same time, the switches still need ISUP attributes carried in SS7 
> IAMs and ACMs for billing, because that's just the information they 
> operate on internally.  SIP-T provides an IP-based way to encapsulate 
> that information.
> 
> SIGTRAN (essentially, SS7-over-IP) is another way to do this.  However, 
> SIP-T is lightweight and easier to deploy.  It also allows you to use 
> existing SIP network elements (proxies, session border controllers, 
> etc.) to route and manage the traffic.   For example, if you were using 
> OpenSIPS + ACC + FreeRADIUS as a CDR catcher, you could run the "SS7" 
> calls between two switches and log the appropriate information as custom 
> attributes.  There are no good open-source implementations for SIGTRAN - 
> nothing as turn-key as Kamailio or OpenSIPS.  SIP is high-level and much 
> easier to deal with and manipulate using a far wider range of tools.
> 
> SIP-T is also becoming an attractive external interconnect option.
> 


-- 
Alex Balashov
Evariste Systems
Web    : http://www.evaristesys.com/
Tel    : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (678) 237-1775