Since its debut in 1995, VoIP has evolved from a hobbyist’s pursuit to a mainstream network technology. Operators around the globe are deploying next-generation VoIP networks. Providers see VoIP and SIP-based networks as an opportunity to cut their operating expenses and boost their bottom line with new revenue.
However, as VoIP and SIP traffic and applications grow, so do the requirements on the underlying network. Operators are discovering that their next-generation networks (NGNs), built on a voice-centric, softswitch architecture, don’t have the scalability and flexibility to support multimedia services, access independence or network growth.
The missing layer
In the rush to get VoIP up and running, many providers took a short cut. They deployed their NGNs as a loose collection of elements interconnected by direct signaling links. Unlike signaling system 7 (SS7) and Internet protocol multimedia subsystem (IMS) networks, there’s no signaling and session layer to handle application-layer tasks.
From a signaling perspective, each NGN element must handle all application-layer related tasks like routing, traffic management, redundancy and service implementation. Every possible route must be defined at each network element, creating a spider web of logically connected SIP nodes.
A better approach is to deploy a SIP signaling router (SSR) – a SIP proxy with enhanced routing capabilities -- to centralize layer-5 SIP routing into the network core. The SSR creates a signaling and session framework that relieves the endpoints of session management duties. The resulting architecture allows the NGN to grow systematically and creates a flexible framework that enables:
- Enhanced application server (AS) selection: The tight coupling between SIP endpoints like SIP phones and SIP application servers creates a challenge for many operators. Any changes made to the physical network, such as adding a new application server, have a direct impact on the way SIP phones access service. The SSR shields the endpoints from changes in the physical network through a process called “abstraction” in which the phones are decoupled from direct knowledge of the complex and changing network; SIP phones just have to be configured with a single abstract address. Endpoints send requests to the SSR, which resolves the address to the appropriate SIP AS platform and routes the request to that platform.
- SIP trunking: Softswitch-based, SIP trunking solutions, which are built on a “per-connection” cost model, can become costly very quickly. And, since softswitches usually are deployed with the switch vendor’s choice of application server, it’s difficult to gain the economy of a “best-of-breed” solution. By implementing an SSR, operators can use a session-based approach to provide fixed-line services to enterprise customers. The SSR routes on-net calls between IP PBXs and off-net calls through a PSTN gateway to local and long-distance fixed numbers. The resulting architecture creates a volume-based cost structure and reduces costs by allowing operators to select “best-of-breed” application servers.
- SIP number portability (NP): For operators with a SIP-trunking infrastructure, performing number portability for VoIP calls can present a particular problem. They can simply “dump” the calls on the PSTN gateway if there’s enough intelligent network capacity and the terminating network is time division multiplexing (TDM). But, if the terminating number is another IP PBX or belongs to a VoIP operator, the call must be shuttled from VoIP to TDM and back to VoIP again. Running pure VoIP calls over the TDM network wastes PSTN gateway capacity and degrades voice quality. Another alternative is to replicate an NP solution in the SIP domain, but that’s a costly approach. Using the SS7 access feature of the SSR, operators can make TDM-NP available to the SIP network. This capability allows the SSR to augment its routing capabilities with data from the SS7 domain. Pure VoIP calls don’t have to be shuttled over the TDM network to perform NP, which maintains voice quality and saves capacity on PSTN gateways.
- Centralized SIP proxy: Expanding the capacity of NGN networks requires the addition of new softswitches. Each new piece of equipment must be provisioned with the routing entries for all of the existing softswitches. The existing softswitches also must be updated with the routing entries for the new equipment. Route management, which is based on pre-defined SIP trunks, becomes increasingly complex as the network expands. Service and subscriber data are tightly coupled with the softswitch, making it difficult to change an existing service or add new applications uniformly. The SSR deployed as a SIP proxy creates a SIP-based reference architecture over the existing network. Calls are routed by default from the softswitch to the SSR. The SSR makes layer-5 SIP routing decisions based on advanced routing algorithms and forwards the request to the appropriate SIP destination.
- Specialized SIP proxy: As operators consolidate their networks, many are discovering that softswitches supplied by different vendors are unable to establish sessions with each other. That’s because each vendor uses a different SIP implementation. As long as an operator deploys equipment from a single vendor, there’s no problem. But, when equipment from another vendor is introduced, interoperability problems arise. The interoperability issue can be resolved with a customized solution, but that’s an expensive route to take. The SSR creates an architectural solution that is independent of the endpoints and eliminates interoperability problems. Deployed at the signaling layer, the SSR serves as a SIP proxy. It routes SIP traffic between the softswitches and serves as a mediation point between them, “fixing” protocol variations on the fly.
History shows that the signaling and session control layer is critical to any large-scale network architecture. Having softswitches and other endpoints perform layer-5 session management may be sufficient for small-scale deployments, but as the network expands, the lack of a capable session framework introduces a host of network issues.
A core framework that offloads signaling and session tasks from the edge elements enables NGNs to expand efficiently and avoids the pitfalls created by a point-to-point, mesh routing. The resulting architecture can expand systematically to support growth, deliver advanced services and create the foundation for future technologies and applications.
Vince Lesch is VP of product marketing at Tekelec