SIP Trunk services are essential bridges in Unified Communications architectures, but provider offers vary widely. Offerings will continue to improve, but it's critical to make sure you understand what's on offer, and what you're getting.
SIP Trunk services bridge carrier-based IP-services with IP PBXs and other IP telephony systems, thereby interconnecting "islands" of Unified Communications (UC) functionality. These services support features like single number service that tie office, mobile and home office phones together. Simply put, SIP Trunk services act as the glue to support multi-location VOIP, IM, web or IP videoconferencing applications. Compared to many legacy carrier access lines and trunks, SIP trunks are more flexible and efficient, and thus can provide part of the hard savings needed to support a UC business case.
When compared to legacy technology, SIP Trunks are relatively new, applicable standards are evolving, and provider offers vary greatly. Thus prospective customers should issue RFIs and RFPs to find the best providers across a wide range of evaluation criteria, including geographic availability; types of access and services that can interconnect with SIP Trunks; vendor interoperability; available features; resiliency and business continuity; design, ordering and provisioning; service monitoring, management and trouble remediation; SLAs; portal functionality; and prices. Since SIP Trunk services play a vital role in the UC infrastructure, I anticipate providers will make significant improvements in their offers, particularly in reliability and functionality, over the next 2-3 years.
A variety of SIP Trunk service providers exist. For this report I interviewed ten facilities-based providers that U.S. businesses most frequently use for legacy PSTN services today. They include AT&T, Global Crossing, Level 3, Orange Business Services, Qwest, Sprint, Time Warner Telecom, Verizon, XO Communications. Several other providers were invited to participate in this research, but did not provide information in time for inclusion in this report.
Salient SIP Trunk Service Attributes
As readers of this section will come to appreciate, a significant degree of variation exists between providers' SIP Trunk offers and related features and services. Based upon the capabilities in current offers and providers' disclosures of planned enhancements, the current degree of offer diversity will continue and even accelerate, at minimum through 2012. Important service attributes include:
* Geographic service availability. With some exception, virtually all SIP Trunk services offered by these providers conform with their DS-1 access footprints--but some limit local SIP Trunk services to the markets in which they offer local services today. Of the ten service providers who participated in this study and now offer SIP Trunks and related services in the U.S., 40% currently offer or plan to make their SIP Trunk offers and related services available outside the United States. For these providers, the number of countries with SIP Trunk service varies markedly--from a low of 10 to a high of 130.
As enterprises with international connectivity requirements know, the availability of intra-country connectivity for existing voice and data services varies markedly by provider; thus--particularly in the current economic climate in which everyone, including providers, faces limitations in capital resources--it's reasonable to anticipate that SIP Trunk availability will remain within a provider's current service footprint, but not exceed it. But as importantly, it is very prudent to verify that a provider's intra-country SIP Trunk availability meets your company's geographic availability requirements.
* Types of access. All providers provision SIP Trunks over DS-1 interfaces, and they all also support SIP Trunk over Ethernet access (which typically requires at least a DS-1 layer 1 interface). These are the only common threads. Provider-specific exceptions include:
--A distinct minority of providers support SIP Trunks over Fractional T1, broadband or fixed wireless access.
--Eighty percent of providers have minimum requirements for SIP Trunks over DS-1 (for instance, at minimum, 384 kbps).
--Ninety percent also have firm maximum thresholds (eg: no more than 40% of provisioned bandwidth, 41 concurrent calls per trunk, etc.); to conform, enterprises whose site-specific real-time connectivity requirements exceed these limits should expect they will need to acquire additional bandwidth (eg: NX DS-1, DS-3, OC-X), because 80% of suppliers' caps are enforced (eg: provider blocks the next call attempt that exceeds its concurrent call or bandwidth limit).
* Service connectivity: All ten service providers now support or plan to support the interconnection of SIP Trunks with MPLS (to support on-net calls), and the same is true of Ethernet services (particularly, VPLS). Seventy percent provide connectivity between their SIP Trunk and dedicated Internet services. As for legacy voice services:
--All support outbound offnet PSTN connectivity through provisioned gateway services, and all support the termination of calls from the PSTN to SIP Trunks (again, through a provisioning arrangement).
--In all cases, offnet PSTN origination and termination extends to both domestic long distance and local calls.
--Local service SIP Trunk availability mirrors the provider’s local access footprint. Providers usually do not extend SIP Trunks to areas where they are not local access providers (eg, FX lines).
--Twenty percent of providers do not terminate PSTN-based toll free services onto SIP Trunks, and thirty percent do not currently offer IP Toll Free services.
* Vendor Interoperability: No enterprise-class service provider relies exclusively on SIP Forum compatibility tests--too many features are still unspecified. To fill this gap, most, but not all providers have developed their own SIP Trunk specifications. However, to control costs, providers will only proactively perform compatibility tests on recent releases of software that support popular brands of SIP-compatible IP PBXs, phones, etc. This of course limits the potential universe of SIP interoperability--today, only 20% of the providers in this study have demonstrated that they have certified more than a handful of vendor products (AT&T is one such provider).
To help alleviate this bottleneck, service providers often make remote testing capabilities available to customers who want to test interoperability on older software releases or with less popular manufacturers. But in these cases, the customer and vendor must directly work with each other on CPE-service interoperability issues--typically with very little, if any, service provider involvement.
Like SIP functionality itself, as providers introduce additional functionality to their SIP Trunk platforms, their proprietary certification tests also evolve. Thus, it’s entirely possible for a vendor’s particular product and release to have been certified for compatibility for all the capabilities and features on a provider’s 2008 platform, but must re-certify as new provider SIP Trunk functions are introduced, or as a vendor issues its newest software release.
* Feature Functionality: The list of individual and group calling features varies markedly. In many cases, features that have historically been supported by SS7 ISUP or TCAP functionality still are supported--only the delivery of the feature over the trunk is based on SIP. Many providers’ SIP Trunks support a subset of their respective PSTN-based features (eg: some, not all N11 features; access to Directory Assistance or lifelines services over SIP Trunks is not universal; etc.). As for other features, here are some important findings:
--The majority of providers have yet to support any SIP Trunk calling features with SIP-only functionality (eg., presence, single number service). In their cases, all of the features supported by their SIP Trunks have legacy PSTN equivalents (eg: call waiting). However, providers plan to make significant headway and introduce a number of IP-dependent features over the next 2 years.
--Providers often group features into categories, as is the case with Level 3's group and individual calling features, which are organized into four different feature packs, (1-3 and Mobile).
--Even on a feature as basic as compression, few providers offer much. For example, 100% support G.711 and G.729a codecs, but only 20% support G.729ab codecs.
--Following up on our earlier discussion on standards, it’s important for customers to verify that the IP PBX vendor provides specific documentation that verifies its release interworks with a provider's SIP Trunk for all the features of interest. For instance, some customers are interested in per-call codec configuration, in which each call can be configured to a specific compression scheme. To work, both the IP PBX/gateway and SIP Trunk interface must support the codecs in question and also support their real-time configuration for incoming and/or outgoing call attempts. In addition, the IP PBX/gateway should have passed all applicable SIP Trunk message flow interoperability tests the provider requires.
* Resiliency. In our examination, most providers' VOIP/SIP-specific infrastructures supplied documentation to bolster their claims that their networks are adequate to handle current volumes, and have a sufficient degree of intra-nodal and inter-nodal diversity and redundancy. But it doesn't hurt to check details very carefully to ensure your expectations will be met. For example:
--Some providers' session border controllers (SBCs) do not currently have any redundancy or diversity, which makes them very weak links in the call handling process.
--Compared to the PSTN, on an end-to-end basis, the VOIP/SIP infrastructures of even major providers handle a fraction (5-10%) of the call handling volumes of their respective legacy Class 5 voice infrastructures, particularly concerning features. Providers plan to grow their VOIP network platforms as demand warrants, but it is possible that some providers will not grow sufficient capacity in advance of demand. Thus when selecting a provider, customers should inquire as to the minimum amount of provisioned idle capacity available at peak periods per provider inter-office trunk.
--Customers should also consider backup power, because in the event of an extended electrical outage, not all providers have the same minimum level of backup power at hand.
* Business continuity considerations. As is the case when considering this issue in relation to a WAN architecture, it’s important to identify the real-time network assets and services that are essential to your company's ongoing business, and to determine the relationship between these assets and profitability. The closer and more direct the relationship, the less your company should rely on one type of service (or alternately, one provider). IT and network engineers frequently focus on the steps necessary to assure a high degree of reliability for intra-company communications. But when using SIP Trunks, they should also carefully consider the necessary redundancy and diversity required to support incoming calls to a main number that ultimately terminates to multiple stations or terminals behind a PBX, IVR or ACD (such calls placed to a main business number or to a toll-free number). Potential alternatives include:
--Continued use of PRI/in-band T1 trunks and related inbound/outbound voice services as backup/overflow for business-critical operations (such as call or contact centers). With one exception, all of the providers who participated in this study do not overflow from SIP Trunks to PRIs or inband DS-1s. They typically overflow from one SIP Trunk to another, whether it is at the same location, or an alternate location. Such providers miss an important opportunity, because most customers will not flash-cut their entire networks from the PSTN to SIP Trunks and site-to-site VOIP. The lack of this feature does one of two things:
(a) serves to retard a customer's adoption of SIP Trunks, or
(b) encourages customers to rely on multiple SIP Trunk providers.
From any single provider's perspective, neither of these alternatives is desirable. In those exceptional cases in which a provider offers SIP Trunk to PSTN backup or overflow, enterprises should verify that the backup/overflow capabilities your company requires function as needed--and get this assurance before signing a SIP Trunk service contract.
--Use of multiple SIP Trunk providers and services, with backup/overflow initiated by the customer. Since not all providers currently offer SIP Trunk overflow, it is especially prudent to consider this option. However, in the U.S., given the diversity of SIP Trunk offers between providers and the lack of many real-time customer-initiated backup/overflow features, this practice is still fairly unusual, and currently is most frequently provided via outsourced offers from large Systems Integrators or non-facilities based MSPs.
As the technology evolves, it could become easier for customers to invoke, and thus become a more common practice in the future. Thus in most instances today, customers who employ multiple SIP Trunk providers directly use both on an active basis, and are only able to redirect traffic from one provider to the other for outgoing calls, not incoming.
--Targeted use of wireless services. Should a SIP Trunk be out of service, employees who are working in an affected office but who have cell phones can use those mobile devices to place outbound calls. But no provider in this study yet offers a hunt group service that on an inbound basis includes termination from a general number to a specific wireless number. In the event of the unavailability of a SIP trunk, only one provider's fixed-mobile convergence (FMC) service automatically diverts an incoming call to a mobile device (Sprint Mobile Integration).
* Design, Ordering, and Provisioning. Although tools exist to automate and interconnect "Day 1" design, ordering and provisioning processes and databases, at this time there are very few standard provisioning intervals for SIP Trunks. For most providers, the process can take 60-90 days or more, and each design is customized.
In only one instance did a provider supply a firm provisioning interval for its three standard configurations; for these, Global Crossing’s interval is 10 business days. But unbounded offer variation and lengthy implementation timeframes cannot scale; recognizing this, several providers informed us of their initiatives to improve these processes in 2010.
But in the near term, it’s important to appreciate the degree of dependency that both providers and customers have on the quality of both vendors and service providers’ resident subject matter experts (SMEs), and on their successful retention of those experts. Customers should not overlook or underestimate this factor when evaluating suppliers.
* Monitoring, Management and Remediation. There is great variance in the typical types of network performance metrics that providers track for these 'Day 2' activities; depending on the provider, they can include some or all of the following: availability, call attempts and completions, post dial delay, congestion, jitter, packet loss, and latency. Only 30% of providers track voice-specific metrics like MOS or R-Factor.
--Since some performance problems recur, retention of performance data is also important. Currently, providers retain performance data from as little as 3 to as long as 14 months.
--Many providers use probes, simulation tools, correlation engines and a variety of other technologies to diagnose problems that can impact the performance of real-time communications, but evidence of speed and accuracy in actual remediation is comparativey sparse.
--Also puzzling to us, no provider specifically mentioned any tools or processes to track and analyze SIP messages, or SIP-related protocols like SIMPLE, or the interactions between SS7 and SIP-based networks.
--We've already examined the important role that provider and vendor SMEs currently play in the Day 1 process--the same is true with Day 2. And as with their design and provisioning processes, virtually all providers plan significant enhancements or upgrades to their monitoring and management capabilities. Given the current state of the industry, this is a work in progress for virtually all providers and vendors at minimum for 2-3 years.
* SLAs. Service Level Agreements lag the functionality of providers' Day 1 design, ordering, provisioning and Day 2 monitoring, management and remediation capabilities.
--Apart from availability, today only 20% supply some types of voice-specific SLAs like call blocking SLAs, or minimal scores for MOS or R-Factor.
--50% provide only performance metrics tied to underlying layer 1 or data networks (chiefly, DS-1 access and MPLS).
--Such paucity in SLA capabilities is at least in part a result of the lack of automation and flow-through in many of the Day 1 and 2 processes described above. Customers of providers' managed SIP Trunk and managed VOIP service offers can obtain more complete SLAs, but given the challenges in providers’ Day 2 management capabilities, they should first verify that these SLAs are supported by adequate application and network management processes and infrastructure.
* Portals. Providers have been pushing portals as the means to interface to their ordering, provisioning, monitoring and billing systems for many years—and all those in our study supply portals for their SIP Trunk customers today, but functionality varies markedly.
--In many cases, only some feature or service changes can be implemented via portals.
--None of the ten providers' portals processes initial SIP Trunk service orders.
--Many providers organize their portal functionality into tiers, and reserve the most aggressive performance analysis tools and reports for users of premium portal services.
--Finally, as discussed earlier, attention to detail is critical. For instance, in examining the online billing capabilities of one provider, I found that it did not supply call detail records (CDRs) for on-net (VOIP-VOIP) calls. Depending upon a particular company’s downstream auditing requirements and the industry-specific reporting and/or regulatory environment it operates in, such an omission could have serious business consequences.
* Pricing. Although it's possible for SIP Trunks to save telecom-services-related expenses, customers should not assume the pricing structure of SIP Trunks and related services is markedly different from the PSTN/PRI world. Many domestic providers currently bill between $1,100-$1,500/SIP Trunk/month. However, some newer providers without a large installed base of legacy PSTN customers and infrastructure offer different pricing elements and simpler pricing schemes. A provider’s SIP trunk-related price elements can include:
--Layer 1 access charges, monthly recurring fees for underlying data network(s) and connections to other network services
--All providers have fees for on-net and/or on-off-net concurrent call paths at peak. Depending on the provider, a concurrent call path can be configured to support only local calls, only long distance calls, or both. Depending on the call jurisdiction, non-discounted fees range from less than $20 to approximately $70/concurrent call path/month. Concurrent call paths must be provisioned on each SIP Trunk and typically are priced per-trunk or per- location. In addition, Verizon Business and XO Communications both support an option that allows for the use of concurrent call paths at peak companywide (eg., across multiple sites served by SIP Trunks).
--Depending on the provider, other common types of SIP Trunk fees include per-minute or message-unit prices for off-net local or long distance calls, prices for groups of DIDs, hunt groups, and other calling features (including use of SIP REFER); portal functionality, one-time or recurring charges for remote local DIDs, trunking or switch diversity, overflow functionality, etc.
--The definition of "on-net" calls varies. For some providers, on-net calls are those that originate and terminate over any form of dedicated access or egress--in-band DS-1, PRI or SIP Trunk. But for most providers, on-net calls are only those that originate and terminate over their SIP Trunks, making it more expensive for enterprises to have hybrid networks (in which some locations use dedicated access to PSTN/PRI-based networks, while others use SIP Trunks). In these cases, calls that flow between one type of location and the other are priced as off-net calls, which usually are twice as expensive as PSTN-based on-net calls.
Conclusion
In many cases today, companies have upgraded and/or replaced legacy PBXs with IP PBXs at select business locations. This provides an all-IP foundation to provide Unified Communications (UC) features to the employees who are served by the new infrastructure, particularly collaborative applications that can yield significant benefits in employee productivity. This can in turn improve the efficiency of important business processes, like improving the responsiveness of a sales organization to a an RFP.
But for companies with multiple locations, the benefits of UC can only be realized if they are able to also participate in an all-IP infrastructure for all intra-company communications (both intra- and inter-site). Today, IP PBXs typically connect to service providers' networks via legacy trunks (typically in-band T1s and T3s, analog lines, ISDN PRIs and/or BRIs). They are IP islands in a sea of legacy technology. By serving as a bridge between customers' IP telephony infrastructures and service providers’ IP networks, SIP Trunks extend UC throughout a multi-location company...even to key business partners, customers, and to remote or mobile employees or contractors.
From the perspective of functionality and reliability, SIP Trunk services are not yet on par with legacy services, so customers should plan on building in a fair amount of resiliency into their SIP-enabled WAN architectures. Business customers often make assumptions about the simplicity or functionality of SIP Trunk services, and expect SIP Trunks to be more mature than most in fact are. But since SIP Trunks are essential to provide UC functionality on a multi-location basis, both vendors and providers have great incentives to substantially improve many SIP Trunk attributes, and thus it is reasonable to anticipate substantial improvements over the next two-three years.
Finally, since current SIP trunk capabilities and planned enhancements vary greatly between providers, it is essential that prospective purchasers conduct a careful and extensive evaluation process.
Lisa Pierce is President of Strategic Networks Group consultancy. This article is abstracted from a larger research document on the same subject. The full article is available from the author.
