Speermint Group Tackles SIP Peering Shortcomings

Imagine you're the CTO of a global enterprise. You've accepted the undeniable reality that communication between your offices comes at a cost--sometimes a hefty one that varies with usage and geographic separation.

Now imagine those interoffice call costs virtually disappearing. Then, imagine your long-distance charges to other enterprises declining. What might be mistaken for an accounting error is actually the result of your astute decision to implement SIP peering. And a standard in the works from the IETF promises to improve communications using SIP peering.

We're hardly strangers to SIP (Session Initiation Protocol) and the dramatic effects it's having on end users, both at home and work. SIP peering is responsible for server-to-server interconnections, providing backbone connectivity between ITSPs (Internet telephony service providers)--or VSPs (voice service providers), in the language of the IETF--and corporate enterprises. These interconnections occur completely in the IP space, using the private or public Internet, never touching the cost-incurring PSTN.Although SIP peering provides some compelling cost and flexibility benefits, the technology has the same drawbacks of VoIP. While it's reliable, backbone peering over a packet-switched network, especially the Internet, is more susceptible to service interruptions than are legacy circuit-switched networks. Furthermore, security between SIP peers requires compatible security technologies; otherwise, calls generally fail or are unsecured. The IETF's SIP peering group, Speermint (Session Peering for Multimedia Interconnect), is working to mitigate these shortcomings.

BIG POTENTIAL

The ideal is to peer all your offices plus as many entities you communicate with as possible, thereby eliminating your dependence on the PSTN and dramatically driving down costs. This is what companies like Kayote Networks, Stealth Communications and XConnect Global Networks hope to facilitate by signing as many members as possible into their SIP "federations," which make all players abide by their rules in order to establish a degree of security and QoS (quality of service) compatibility.For enterprises or providers that cannot justify joining a federation, Speermint provides a series of BKMs (Best Known Methods) that outline methodologies for handling real-time sessions over peered networks.

The overarching purpose of the Speermint group is to extend the SIP protocol with the development of various architectures designed for specific use-case scenarios. All architectures are constructed to provide identification, signaling and routing of real-time and delay-sensitive communication while establishing and maintaining trust, security and resistance to abuse and attack.

To operate truly independent of network type, the group has set its focus on the application layers--Layer 5 and above. This lets it generate models that are generic enough to satisfy its efforts independent of the underlying network, be it a DSL connection or OC-48. It may be an overly optimistic assumption at this point since there are considerable differences--such as QoS capabilities--between those two technologies. The group acknowledges that fact and may consider rechartering in the future to encapsulate QoS and traffic-engineering mechanisms.

At the time of this writing, the Speermint group had released five Internet Drafts that outline some of the basic principles of SIP peering. In essence, these drafts extend the established SIP standards to accomplish Speermint's objectives. Of note, two of the drafts outline a logical grouping of peering functions and subsequent message-flow phases.

The Internet Draft related to message-flow phases introduces discovery, policy and security phases to a SIP session, consisting of exchanges and agreements between peers before the communication of SIP signals and call setup. These additional sequences are the first steps toward establishing dynamic policies configurable by system administrators amongst multiple SIP peers.Although Speermint could offer the same level of protection and reliability as that available in federations, don't expect providers and enterprises to create their own peering communities. The private infrastructure offered by federations such as the VPF provides an extra layer of security and performance that public networks simply cannot offer. However, do expect to see some large "Speermint-enabled" open peering communities pop up in the years to come that the rest of us can join.

Matt Vlasach is CEO of Pacific Swell Networks. Write to him at [email protected]