Ask the Analyst: SIP TrunksAsk the Analyst: SIP Trunks

From E911 to MPLS to local number portability and DID issues, plus migration planning for the future, the leading enterprise carrier analyst answers end users' questions about SIP Trunks in the real world.

In our latest Enterprise Connect webinar on SIP Trunking, we got more than 60 questions from the audience of 500 people--way more questions than our experts could possibly answer in the time allotted. So our analyst speaker on that webinar, Lisa Pierce, president of Strategic Networks Group, has agreed to pick out some of the most provocative questions that didn't get addressed during the webinar, and answer them here.

To watch a replay of the SIP Trunking webinar, click here

Q: Today, I have the flexibility to route my DID calls between two different Verizon central offices, if I replace the PRIs with SIP trunks, both Verizon and AT&T do not offer CO redundancy. How do I protect my DID calls from a CO outage?

A: Many SIP Trunk service providers offer customers optional alternate destination/re-routing features for inbound calls (from the primary SIP trunk to a secondary SIP trunk, some also offer SIP to PRI overflow/backup). Some also extend the same functionality to DID calls. Depending on the carrier and the particular backup/overflow feature, some are implemented via service order, some may be implemented ‘on the fly’ (via the provider's portal). Compared to the number of Class 4 or 5 Central Offices, carriers have comparably few nodes with all of the intelligence and functionality that VOIP requires. Thus most carrier "VOIP" nodes are architected with intra-node redundancy (in addition to inter-node redundancy). This is an example where CO survivability can be managed more easily with IP than with legacy technology--but there are still issues. SIP Trunk survivability is essential. SIP Trunk features vary greatly between providers, even on backup/overflow, so customers should look at a wide swath of providers, not just those currently employed.

Q: The backup capability: Is it provided by the carrier as part of the offering or is it usually provided by another carrier for a separate price?

A: The SIP Trunk provider offers the backup/overflow feature(s) as "options", so these are in addition to the base price. You can also employ a second facilities-based SIP Trunk service provider, but be prepared to invest the appropriate amount of time doing your homework (CPE-primary SIP trunk, CPE-secondary SIP trunk), and getting both providers’ services to operate as you expect. Using two SIP Trunk providers may require additional CPE resources at the location where you terminate your SIP trunks. Please note this type of survivability will only work for outbound calls and can also work for toll free calls, but will not work for DID calls. I'm currently aware of only one domestic provider that offers a network re-routing feature that overflows to another carrier's phone number for incoming (non toll-free) calls (that carrier is XO).

Q: What about E911? Does every carrier offer this or do you have to find another provider for this service?

A: Every domestic SIP Trunk provider I've evaluated offers e911, but they all aren't the same. Some with large legacy customer bases will use their TDM-based e911 services. Some smaller providers wholesale e911 services from other providers (so it's important to verify who is actually providing the service). Most important is the treatment of far-end users. Consider a small branch office in Wichita with remote IPT functionality connected by MPLS to the Dallas IP PBX (HQ), and the IP PBX has a SIP Trunk connection to the carrier. To which e911 dispatch center will an e911 call originating in Wichita and going out the SIP Trunk in Dallas route-- Dallas or Wichita? What originating number/location will the e911 dispatcher pull up? We know what should occur, but different carriers handle this differently. Carriers also differ on registration of remote users (and updates to that information).

Q: In PRI environment we use Erlang study and a fairly common ratio of 4 Users per Trunk. What is a common or estimated ratio of Users/Endpoints per concurrent call/SIP trunk?

A: In the webinar, we briefly discussed re-architecting both real time and non-real time applications around the data center. There are a number of ways to estimate the amount of voice traffic transiting the data center for intra-company and inter-company calls--ideally companies would have at least two sources of data on per-site call attempts/completions (PBX and carrier data), and compare called party number fields to identify intra vs. inter-company volumes. There are a number of ways to obtain this kind of data; suffice it to say that it's always better to know than to guess, because one way or another, guessing wrong costs money. Many companies still use the 10:1 ratio from legacy trunks (non-contact center application); the ratio supplied by the questioner is considerably lower, so I wonder about the applications that company is running. Also, since more people make/receive calls on mobile devices than in the past, it's possible that the company can put more endpoints on a concurrent call path than would have historically been possible; if so, even greater savings are possible.

Q: Wouldn't an SBC be more often the responsibility/need of the SIP trunk carrier (seems it would be more flexible if that was the last touch point "in the cloud" prior to delivering service to an enterprise/CPE)?

A: SBCs are necessary at the carrier end of the SIP Trunk, absolutely. There are a number of times when similar functionality is essential at the enterprise end. For example, if a company employs an IP PBX that has not passed the carrier's SIP Trunk interoperability tests, an SBC that has passed the tests can interconnect the IP PBX and the SIP Trunk. Premises-based SBCs perform a number of functions that carriers haven't historically performed as part of a basic access service. A few examples include termination of hosted IPT/UC/contact center services at the company site (can be bundled into the hosted service's price), H323:SIP interworking, implementing company-specific security policies for real-time communications, and authenticating remote access users who access premises-based UC/IPT functionality via SIP Trunks/Internet service.

Q: I have a nationwide customer looking to consolidate thousands of local LEC connections into central SIP trunks. What are some of the LNP (local number portability) challenges you've seen and how are they being overcome?

A: LNP "challenges" typically are administrative/operational in nature and don't directly impact SIP Trunks. LNP problems are further back in carrier service logic--getting the "old" terminating carrier to appropriately update records that are shared with all carriers; updated records point the terminating traffic to the new carrier. LNP does not directly point traffic to a specific customer's access line/trunk in a particular provider's network--that is determined by the terminating provider after the it receives the call initiation message (in SS7, an ISUP Initial Address Message and if SIP is employed on an inter-carrier basis, an INVITE message) from the next to the last (N-1) carrier in the call flow sequence. However, if a company is considering implementing SIP-based local/Virtual Telephone Numbers as the question (partially) implies, functionality (and pricing) of this feature varies by provider. For example, some providers will offer VTN functionality for incoming calls, but on an outbound basis do not transmit the applicable VTN to callers (for outbound calls to a particular local market). And not all SIP Trunk providers offer VTNs. These are good examples of why digging in and doing your homework is so important.

Q: Providers will not admit this but can you confirm that they are downsizing their TDM equipment now as they increase SIP?

A: The good news and bad news of migrating to new technology requires that R&D, capital and operational spending increase on new technology, and level off and then decrease on the old technology. No one can stay in business if they didn’t ‘manage this--witness the "migration" from ATM and Frame Relay to Internet VPN, MPLS and Ethernet services. In the case of TDM/PSTN, reductions in both capex and opex have occurred for many years (just think about the tens of thousands of LEC union layoffs over the last 10+ years). It's been happening in a semi-controlled fashion. I say that because in some cases I don't think the carrier anticipated the full impact of the layoff/"incented" early retirement in advance. The level of continued investment in legacy technology will vary by provider; some of the carriers that say their continued PSTN investments are sufficient for service quality to be unaffected will have trouble backing that up, because there are a fair (and growing) number of instances where real-world results (and data provided to state regulators) say otherwise. For instance, the State of West Virginia recently required that the ILEC put up a reserve fund of over $74M for copper wire maintenance (the ILEC wants to exit the West Virginia market).

Q: Will providers or have they already do away with the long distance price model? For instance--all calls within your home state considered local calls?

A: This will depend on the provider. Smaller providers have great incentives to keep it simple to win new customers, and many do. Large incumbents could be tempted to continue offering prices that are based on longstanding regulatory boundaries (Inter-vs. intra-state; intra-state inter LATA vs. intra state-intra-LATA). With the adoption of digital TDM Class 5 COs in the late 1980s/early 1990s, a strong argument then existed that intra-state distance-based pricing had little basis in real (underlying) costs. But these price disparities persisted nonetheless. At the least, I think large incumbents will provide more user-friendly prices in areas where they are not the legacy ILEC (e.g., extending their local service areas to new markets via VOIP/SIP).

Q: How can SIP trunks eliminate the use of analog lines?

A: This will depend on the alternate architecture chosen (the webinar highlighted a data-center centric model, which favors the use of larger digital trunks to support calls ultimately going to/from many company sites, vs. individual analog lines going directly from those sites to the PSTN). As discussed then, if an organization uses SIP Trunks for inter-company calls, then the organization can reduce or eliminate the number of analog lines it employs. Why? Most providers' SIP Trunks are capable of supporting both LD and local calls, on both an outgoing and incoming basis. Some customers may want their small locations to directly access the carrier network (eg. not traverse the data center). Most carriers offer SIP Trunk interfaces on fractional T1 access (remainder of bandwidth can be used for data applications), some also support SIP Trunk interfaces on broadband access. And more will, going forward. For instance, Cable Labs is working on this for an extension to the Packet Cable 2.0 specification.

Q: Can SIP run over an MPLS circuit?

A: Yes, many companies use SIP on MPLS to control intra-company calls/sessions between IPT-equipped locations. This CPE-based SIP (typically heavily proprietary) should be part of the flow of real time traffic on those MPLS ports (eg, part of Voice CoS). Companies that employ SIP this way still have to determine how their IP PBXs face the external world and even remote users (to support inter-company IPT/UC applications). Carriers’ SIP trunk services provide that (external, inter-company) connectivity. Depending on a company’s business and its voice and data WAN architecture, it's not uncommon for 60-90% of incoming/outgoing calls to be "offnet" (e.g. would use a SIP Trunk service). Assuming a company uses the same carrier for MPLS and SIP Trunk services, the company may elect to employ the carrier’s SIP Trunks at multiple company sites (for intra-company voice/UC survivability).

Q: Are SIP Trunks used internationally?

A: Yes, but availability varies by provider, and is a moving target. Some examples are illustrative--at this time, both AT&T's and Verizon's SIP Trunk services are available in other countries, but their international SIP Trunk footprints are not as broad as legacy access (PRI, E1). Compared to either carrier, Global Crossing has a much larger SIP Trunk services footprint today (overseas). Another example is Orange Business Services, which provided only H.323 access for many years (for VOIP), but now also offers SIP Trunk Services (in all 105 countries where it also offers H.323).

Q: How long will it take for SIP Trunks to replace traditional trunks?

A: On a per-customer basis, it can be as short as a matter of months. Many new carriers don't offer traditional types of access--they bet the farm on SIP. Carriers of any size, with a fairly large installed base of traditional fixed access lines and trunks, are looking at a 5-10+ year migration of the installed base. But at some point not too far down the line, they will also begin to (1) actively encourage migration to SIP (typically this commences by raising the price of legacy access, something already underway), and (2) subsequently require that (all) new installations be on SIP Trunks (obviously some carrier services require the provisioning of SIP Trunks today). Vendors of legacy access equipment aren’t churning it out at nearly the rate they had even 3 years ago. How long for every single legacy access line or trunk on the planet to bite the dust? My guess is 15-17 years, except for some persistent special applications (eg, possibly some government/military contingency applications).