HIGHWAY SIP REVISITED

It would be easy to blame the economy for the brick wall many technologies have hit on their road to next generation glory. But sometimes, despite an aggressive industry push and a healthy dose of venture capital, a technology is just not ready for prime time — at least not the way most telcos define it.

For a more detailed look at what survey respondents said about SIP and VASA's suggestions for improvement, see the table below.

That is the conclusion drawn by VASA about Session Initiation Protocol (SIP) after analyzing responses to a request for information (RFI) sent to various members of the SIP community last summer. VASA compiled the data and issued a report in November that suggested vendors go back to their labs and complete the job. To the possible chagrin of some in that community, VASA's Rebekah Keating will present the group's finding this week at the IN/IP World Forum in Orlando.

VASA is not anti-SIP, Keating said. “We were like everyone else in the industry: We thought SIP was ready for prime time and that's what we expected to find out,” she said. “So we were very surprised by the results of the RFI.”

VASA is composed largely of members of the previous IN Forum (not to be confused with the IEC's IN/IP Forum), a consortium of technologists focused on the Advanced Intelligent Network model of telecommunications. Today, it is a technology-independent group that assesses global technologies. The organization sent the RFI to vendors, service providers and carriers, mobile operators and others in the industry hoping to gauge the true state of the protocol's interoperability, quality, scalability and security. The survey also hoped to identify new services and get a real-world snapshot of current and planned deployments.

The results showed that, according to survey respondents, “there is little reason for a carrier to immediately deploy SIP on a large-scale basis.”

VASA's request solicited information in four categories. A general list of questions asked for information about participants' overall view of SIP's role in the network (whether it was primary or complementary to the existing architecture), of its promise and its limitations, and about current and planned deployments. The other categories were: architecture, which dealt in part with back-office interfaces, security and end user devices; services and applications, which examined interworking with the public network, service architecture and ancillary systems such as local number portability; and operations, which looked at service creation, interoperability and manageability.

At the scale required by large carriers, SIP failed in all categories, according to the responses received by VASA. The technology's one saving grace appears to be its ability to deliver presence-based applications. “We just called it as we saw it,” Keating said. “Our paper doesn't say we should drop everything we are doing with SIP and throw it away. We are simply saying, ‘This is where you guys need to be working to make it deployable.’”

The RFI was developed with input from 10 service providers and network operators — six of which were U.S.-based. VASA received in-depth responses from 23 companies (mostly vendors) and comments or presentations from Equant, SBC Communications and Telecom Italia.

Verizon Communications — which was instrumental in formulating the RFI through Jim Haley, VASA's chairman and distinguished member of Verizon's technical staff — declined to comment on the status of SIP. Haley said Verizon was not ready at the time to respond. Speaking for VASA, Haley said there was no agenda in developing the RFI. “VASA is neither pro-SIP nor anti-SIP. We just wanted to see what is really out there, and based on what we got back, it was successful.”

VASA succeeded in three of its six objectives. The organization was able to identify generally available SIP-enabled products and solutions; assess SIP's success in addressing network requirements; and understand how SIP could complement existing technologies. Partly for competitive reasons, the RFI failed to get a snapshot of existing SIP deployments or catalog planned deployments. It also failed to identify any particular applications or architectures that would make SIP a must-have technology — primarily because, according to respondents, there aren't any.

“We thought that once you had SIP, there would be all these wonderful services, but from what respondents said, there don't seem to be a lot of new services one could lay on SIP that you can't already do with [advanced intelligent networking,]” Haley said.

However, based on what some who chose not to respond to the RFI said when interviewed, the RFI was flawed. “It was edited by Bell-heads that look at telephony from the old [public network] model,” said Henry Sinnreich, distinguished member of engineering for WorldCom. “Obviously, the Internet has different dimensions, which they have not considered in the RFI.”

Both Sinnreich and Nortel Networks' David Sliter, general manager of Nortel's interactive multimedia server product line, say SIP has delivered on its promises. WorldCom's flagship IP-based communications product, WorldCom Connection, is based on SIP.

“We are absolutely convinced that SIP is ready for prime time,” Sliter said. He said Nortel did not respond to the RFI because it never received one.

Sinnreich said he advised WorldCom not to reply because the RFI was formulated in such way that only an analogy to the legacy PBX was permitted. WorldCom's view is that the RFI didn't meet its business model for SIP services, which are more network-based and less reliant on the PBX or public network.

“With all respect, you cannot equate an RFI, which is a piece of paper, with real hands-on experimentation across the world,” Sinnreich said. “I don't want to criticize this report specifically, but any paper study not based on numbers from real trials on the public Internet is not worth the paper it is printed on.”

Within VASA, however, members have said the report is invaluable in discerning the mixed messages they have received about SIP over the last year. In no order of criticality, the report identified common issues even among SIP proponents and dispelled misconceptions in areas of security, scalability, services, standardization, interoperability, adherence to regulatory mandates and quality of service.

VASA did not rank the issues because it wanted to maintain the overall industry perspective and allow members to determine which items were most important to them, Haley said. However, the perceived lack of scalability was most surprising to VASA and, besides respondents' inability to identify a valid business case for SIP deployment based on the lack of new services and unproved savings in operational expense, security seemed to be foremost on their minds, followed by interoperability.

According to Keating, security is the only real show-stopper, but all aspects of SIP need to be addressed. Many of the issues overlap, such as security and standardization.

The very thing that makes SIP so powerful on paper — its flexibility in allowing third-party extensions to the protocol for developing new services — is what concerns those who are more used to the closed, secure world of signaling system 7. Often, SIP extensions developed by third parties are proprietary and do not interoperate with standards-based solutions. What's more, they open signaling and call control to the Internet.

Nortel, for one, recognizes the problems created by proprietary extensions and makes an effort to bring its extensions back into the IETF for consideration as part of the standard. “Some [developers] don't bring them back into the standards bodies and that can be a problem, but those are few and far between,” Sliter said.

Nortel has come up with its own extensions to address the very issues raised through the report about network address translation (NAT) and firewalls, but the company says it has published them back into the IETF. Extensions, though, are only part of the security problem (or the solution depending on your point of view.)

VASA's report indicates that more work must be done to develop a standard framework for basic authentication and privacy, and that respondents are concerned about the interception of signaling and media traffic, the introduction of malicious traffic through open interfaces, authenticating endpoints and securing management interfaces based on open standards. There are also concerns about the problems of NAT and firewalling that are inherent in IP networks.

“I don't want to totally second guess the report,” Sliter said. “But I don't understand some of the contentions. Our experience suggests that openness and interoperability are absolutely ready.”

Nortel has added SIP support for its Succession softswitches and interactive multimedia servers to support direct packet interworking with SIP-based PBXs and gateways. It is in trials with Bell Canada in anticipation of a large-scale deployment.

Sliter conceded the lack of new services, but argued that when it comes to services, it is all about the bundle. “If you look at instant messaging on its own, or even presence and voice over IP, they are interesting but not unique,” he said. “But tying them into a bundle and allowing users to control their communication rather than having it control them is unique.”

The banter arising from such a report will only make those in the SIP community more determined to prove its viability. And when that community is lead by telcos, that's just what it will take.

So far, though, “they haven't proven it with SIP or IP, period,” Keating said.

To that, Sinnreich said: “If you want to see how Internet telephony works, just use it. I only regret that they can't call me on a SIP phone; we could clear this up in five minutes.”

ACCORDING TO RESPONDENTS...

SIP deployments Disparity in projected timetable for implementation Inconsistent standardization Extensions lead to interoperability issues Mixed messages on the meaning of deployed Few firm plans to deploy SIP in the next two years Government mandates Just because IP Is not regulated doesn’t mean it won’t be Uncertainty about SIP’s role as an overlay technology leads to uncertainty its regulatory requirements Vendors deem number portability as a network issue No solutions for lawful intercept Ditto for emergency services Security NAT and firewall issues limit the deployment of SIP No standard framework for authentication and privacy SIP throws the door open to signaling and media traffic Service Creation SIP does not make service creation any easier Standards and tools for SC are immature Service interaction has not been addressed SIP will not shorten time for introducing new services Billing and other OSS interworking has not been thought out

 

CALLS FOR ACTION...

SIP deployments Provide interoperability test results to carriers Demonstrate financial success with greenfield deployment Provide migration models Provide end user feedback regarding SIP experience Demonstrate successful management of feature interaction Government mandates Address issues of number mobility, intercept, and E-911 now -- not when required by law. Security Make primary during development of third party service creation Service Creation Increase focus on feature interaction Provide mechanism for controlling interaction of services from different sources Scalability Agree on appropriate metrics for measuring sclability Address interrelationship of scale, performance, and reliability of SIP-based networks and applications Quality of Service Develop tools and measurements for non-voice or semi-real time services such as IM.