Introduction
SIP trunks have proven to drive costs savings, boost flexibility and address capacity challenges, but how do you ensure that your designs will deliver the high availability needs of your enterprise? Before completely retiring legacy TDM trunks, consider the benefits of blending them with your SIP trunks and taking advantage of technology diversity to drive higher availability. If your requirements demand ultrahigh availability, then take it a step further and consider carrier diversity.
In this article, we will cover a number of designs that blend the use of new and traditional technologies as well as multiple carriers to deliver highly available solutions for your voice environment. Regardless of your voice needs, technology and carrier diversity can provide simple and often cost effective ways to guard against downtime of voice services.
SIP Trunk Deployment Using Site Diversity
In a simple SIP trunk deployment, a carrier will provide inbound and outbound voice services over an IP Wide Area Network (WAN) connection (Typically MPLS or Internet). There are two deployment models in this case. The option you choose depends entirely on the carrier you procure your SIP trunk service from.
If you obtain SIP trunk services from your existing WAN MPLS vendor then you can simply consolidate voice and data services on the same physical links. This option produces some of the most attractive financial benefits, since you eliminate separate physical voice and data circuits. If you instead use different providers for SIP trunk and WAN services, then you'll need to deliver your voice service over-the-top of your WAN. In this case you need to provision separate physical links to your SIP trunk provider.
Regardless of your deployment strategy, one of the main advantages of using SIP trunking is the ability to terminate these services in multiple data centers. With this approach you provide a very simple, cost-effective and highly available solution.
In the following diagram you can see how voice services from a single carrier are delivered to multiple data centers to create high availability through site diversity.
With this design, it's also important to split the call control function across the two data centers and maintain heartbeat via a back channel (in this example Metro Ethernet). This ensures that both call control and carrier services remain available in the event of a single data center failure.
External or off-net calls are delivered to and from remote offices via the data center trunks by routing this traffic over-the-top of the second carrier MPLS WAN. Enterprise Session Border Controllers (E-SBCs) provide the enterprise security boundary between you and the voice service provider, while carrier Session Board Controllers (SBCs) provide the security boundary between you and other enterprises. E-SBCs, in this case, also provide additional benefits like addressing fax issues, manipulating SIP headers, and protocol interworking. See my September article titled "Enterprise SBCs: Why they're so Important" for additional information on the value of E-SBCs.
Next page: SIP Trunk and PRI Technology Diversity
SIP Trunk and PRI Technology Diversity
Using SIP trunks in a site diversity deployment provides a high level of availability but does not protect against IP routing or SIP protocol issues that may occur within the voice service provider's network. To address this issue, consider retaining some of your long distance voice PRIs in your data centers to provide technology diversity. If your data centers happen to be in the same area or LATA as your large offices, then maintaining local PRIs also provides additional benefits. For larger sites that are not in the same LATA, keep one or a few local PRIs in place where voice services are critical.
In the following diagram we have incorporated voice PRIs and SIP trunks to deliver site and technology diversity in our design:
In this case, we maintain both local and LD PRIs in the data centers to protect against SIP service or protocol fails. In a SIP service failure, LD PRIs will be used to route long distance voice traffic for the majority of U.S. locations, while LEC PRIs will carry local voice traffic for offices in the same LATA as the data center(s). Larger offices not in the same LATA will have local voice traffic route out the LEC PRI(s) local to their site while long distance traffic will route out of the LD PRIs in the central data centers. This is accomplished by configuring the SIP and PRI trunks in the same trunk group, with the SIP trunks configured as the first resource used, and the PRIs as secondary. When configured this way, the IP-PBX will reroute calls out of the PRIs when the SIP trunks are not available.
Inbound call handling for DIDs is a bit more complex. The best approach for managing this today is to use one of the newer features offered by SIP trunk providers. Features differ by provider, but include DID failover, DID Forwarding or TDM failover.
Each of these options will reroute your inbound voice calls if the SIP service becomes unavailable. While all of these options address rerouting inbound DIDs, the capabilities can vary dramatically from carrier to carrier. Some require manual intervention to trigger the failover while others failover automatically. Some preserve calling party information and others don't. Make sure to ask what is and isn't available when the service reroutes. Also make sure you fully understand what the triggers are for options that failover automatically. Some may simply ping the IP address of your E-SBC and only failover if there is a loss of IP connectivity, while not being able to detect loss of SIP services.
An additional benefit of this design is capacity management and overflow. Larger site and data center PRIs can be used to support outbound traffic during times of high enterprise call volume. This is critical to ensure that enough capacity is reserved to handle inbound calls for DIDs that have been ported to the SIP service provider.
Regardless of the options you choose, make sure to review the features and capabilities extensively with your provider before committing to and provisioning one of these services.
Next page: Multi-Site Contact Center with Technology and Geographic Diversity
Multi-Site Contact Center with Technology and Geographic Diversity
Taking advantage of technology diversity should also be a key design consideration when using SIP trunks in a contact center. The majority of contact centers today use toll-free services for inbound calls, and one of the benefits of toll-free services is geographic flexibility. Toll-free numbers using TDM services are delivered over LD trunks and don't have the geographic or LATA restrictions that DIDs have with legacy TDM services in the U.S.
In this case, you have the ability to deliver inbound calls from the same toll-free number(s) to multiple locations using PRIs, DS3s or other legacy TDM options. In turn, you can blend these options with SIP trunks and deploy to multiple data center locations. This results in a highly available solution by using both technology and geographic diversity. In addition, by keeping these service terminations in your hardened data centers, you deliver the added value of power and environmental controls.
It's important to note that these examples are U.S. centric. Restrictions exist on the use of toll-free or free calling services outside of the U.S. There are options for international toll-free, but deployment and usage options will vary depending on country and call routing requirements.
Many carriers that provide toll-free services offer additional advanced features, including percentage allocators. Percentage allocators deliver inbound voice traffic for a single toll-free number to multiple trunk groups. As the name implies, you are able to specify the percentage of traffic allocated to each trunk group. Carriers offering IP toll-free and percentage allocators have the ability to spread calls across a combination of TDM and SIP trunk groups as a percentage of total call volume.
With this design, not only can you maintain active traffic flows across all your trunks, but you can also place trunk groups in and out of service as needed. A trunk group can be taken out of service by simply setting the allocated percentage to zero. This also gives you the ability to control the introduction of new trunk groups into your environment. As new trunk groups are added you can control the introduction by starting with only a small allocated percentage of traffic.
Percentage allocators are a great complementary service for geographic and technology diverse designs. As with DID failover options, make sure to review the features and capabilities extensively with your provider before committing to and provisioning this service. Many of the basic percentage allocation services don't detect trunk or trunk group failures and may continue to route traffic to a failed path. Your provider may require you to purchase additional services if you want automatic traffic rerouting in the case of failure. Take the time to do detailed mapping of your call flows and failure scenarios and conduct thorough design reviews with your service provider architecture and engineering teams.
Finally, as a general guideline, each data center should be sized to handle the average busy hour call volume plus 50%. This allows a single data center to support normal call volumes plus failover surges in the event of a data center failure.
Multi-Site Contact Centers with Carrier Diversity
Despite all best efforts, carriers do experience toll-free service disruptions in their networks as a result of abnormally high call volume, configuration error or natural disaster. Enterprises requiring ultrahigh availability in their contact centers should consider carrier diversity. Carrier diversity is used to drive ultrahigh availability in other applications, such as Internet access, and the same benefits apply to voice services. Sorell Slaymaker said it best in his May 2010 post titled, "RespOrg: Toll-Free Carrier Diversity," which read, "It is advantageous to use multiple carriers to deliver toll-free calls. Most enterprises use multiple ISPs for Internet connectivity to increase reliability, disaster recovery, and control costs; the same model should apply for toll-free voice traffic".
One of the best approaches when implementing carrier diversity is to use a Responsible Organization, RespOrg to direct your toll-free number(s) to multiple carriers. A RespOrg is a company that has access to maintain and manage toll-fee numbers within the SS7 network via the Service Management System. Every ten-digit toll-free number is managed by a RespOrg. While RespOrgs are typically telephone companies, there are a number of independent service providers who are RespOrgs.
A large enterprise that wishes to have carrier diversity can use an independent RespOrg to direct calls for individual toll-free numbers to multiple carriers. Independent RespOrgs have no direct affiliation with carriers and don't profit directly from carrier transport services. As a result, they have no financial incentive to keep your traffic on a single vendor network like carriers do. In addition to being able to designate which carriers to route calls to, an independent RespOrg can configure other rules that drive additional cost savings through least cost routing based on call origination, time of day, day of week, etc.
It is also possible for an enterprise to become a RespOrg. This process is administered and managed by the Bell Operating Companies (Verizon Communications Inc., AT&T Inc. and Qwest /CenturyLink Inc.). The process is relatively straightforward and can be found on the SMS/800 website.
Summary
SIP trunk services continue to mature and gain traction in the industry. SIP trunks have proven to deliver cost savings, flexible capacity and geographic freedom when blended with legacy TDM trunks, delivering high availability though technology diversity. For ultrahigh availability requirements, implement carrier diversity and use or become an independent RespOrg.
Jim Allen is an independent consultant who will lead multiple sessions on SIP Trunking at Enterprise Connect Orlando 2014