Which technology will dominate the future of IP communications?
By Richard Grigonis
Session Border Controllers (SBCs) are devices that, among other things, allow VoIP calls to penetrate firewalls and other "walled-off" areas of the Internet and other IP networks. Recently there has been an ongoing, heated debate over whether SBCs will survive in the face of emerging softswitch (also spelled "soft switch") technology that's capable of performing many (if not all) of an SBC's functions. SBCs may simply keep evolving, avoiding extinction entirely.
Sensing that we could provide a public forum in VON Magazine for informed, intelligent discussion每and not wanting to miss out on the possibility of sparking an entertaining, passionate exchange of ideas, Yours Truly asked six companies involved in various aspects of the debate to participate in a sort of "Delphi poll" that proceeded as follows: Each company provided an initial position statement. All statements were gathered into a single file and this was sent back to all the companies along with a note informing them that they had the chance to reformulate their statements based upon the other statements' contents. It is this final "second round" of position statements that are reproduced below. We hope you find them enlightening. Your comments: readers@vonmag.com.
Integration Rules!
Products in the "Session Border Controller" category have grown to include features such as far-end NAT traversal, near-end NAT traversal, inter-carrier peering, intrusion prevention, protocol repair, QoS, call recording / intercept, protocol transcoding, encryption transcoding and more. When someone asks, "is there a need for a standalone SBC, or should that functionality be integrated somewhere else," the response inevitably becomes, "it depends on what you need to use an SBC for."
Existing candidates for integrating SBC functionality include: firewalls, softswitches, communications servers, routers, gateways and access servers. Arguments for integrating SBC functionality can be made for each of these components. In fact, vendors in each of the aforementioned categories are currently evaluating the impact of adding SBC functionality. However, it is still necessary to specify the exact functionality as different network elements will not need to support all possible SBC functions.
Our view is that much of the existing SBC functionality will ultimately end up integrated into other network elements. As a vendor of both SBC products and SBC technologies, we view this as a positive change allowing us to not only provide a standalone appliance, but to also provide enabling technology. However, regardless of the move to support SBC functionality throughout different networks components, the demand for standalone SBCs remains healthy and continues to grow.
〞Joel Fisher, vice president of business development, Jasomi Networks.
Mutual Coexistence
SBCs perform several distinct roles, including security, NAT traversal, topology hiding, protocol translation, and accounting. Some of these, in particular those to do with the basics of handling calls (such as translating protocol variants, accounting and lawful interception of call details) are logically part of the softswitch每their incorporation into SBCs today is, while often necessary, simply tactical to address incompatibilities between, and deficiencies in, some softswitches.
Other features, in particular those which have to do with protecting the softswitch and core IP network from external attack, are essentially a "smart VoIP firewall" that may survive as a standalone entity or become subsumed into the edge router每in just the same way that HTTP firewalls are a different network element than web servers. Moving security policing to the edge allows service providers to optimize networks by minimizing backhauling of voice traffic while meeting regulatory, security and billing requirements.
MetaSwitch's approach is to support a rich set of protocols (including many variants of MGCP, SIP and H.248) and call handling capabilities (including billing and CALEA), while working with SBC partners to deliver maximum security for open IP networks, and driving down the overall cost of VoIP solutions.
〞Andy Randall, vice president of marketing, Metaswitch.
SBCs Not So Easily Mimicked
Some people argue that softswitches eventually will subsume the functionality of Session Border Controllers (SBCs). However, softswitches and SBCs are significantly different in functional focus and core technology. Moreover, differing locations in the network architecture precludes most of the SBC functionality from "collapsing" into the softswitch.
Softswitches and SBCs are completely different core technologies:
SBCs are packet processing networking devices whereas soft switches are database / transaction processing server software.
SBCs process both media (voice, video, fax) and signaling streams whereas softswitches only deal with signaling.
SBCs perform packet classification, measurement and manipulation from layer 2 through layer 7; softswitches focus only on application layer (layer 7) transaction processing.
These differences mean that softswitches are inherently ill suited in mimicking most SBC functionality. Similar arguments were made with web security products and routers. Eventually, some security functions were subsumed into the routers, but the security products are nevertheless a multi-billion dollar stand-alone business.
〞Jim Greenway, vice president of marketing, Kagoor Networks.
SBC Envy!
Let's face it, the softswitch vendors have session border controller (SBC) envy! It's economic envy rooted in the fact that SBCs have become strategic to service providers in the delivery of secure, high-quality IP interactive communications每voice, video and multimedia sessions 每 across IP network borders.
Integrating signaling and media control, SBCs perform critical inter-network mediation and control functions:
Addresses每connectivity & security (topology hiding, NAT traversal, address translations)
Signaling rate每overload & DoS protection for softswitch
Bandwidth & QoS availability每admission control
QoS每marking/mapping and routing control
Signaling protocol每interworking and basic interoperability
Codec每normalization via stripping, or routing to network or transcoder
Other translations每DTMF in-band vs. out-of-band, signaling response codes
Not burdened by the heavy cost of DSPs from a financial and latency perspective, an SBC is extremely economical, compared to a media gateway, at the border between two IP networks.
As the leader in this category, Acme Packet can tell you that 11 of the top 25 service providers in the world are already deploying standalone SBCs. All the rest are evaluating and will make decisions within 12 months They all intelligently recognize that SBCs for IP-IP network borders, media gateways for IP-PSTN network borders and softswitches have no overlap. They each perform a very different role with very different economics.
〞Jim Hourihan, vice president, marketing and product management, Acme Packet.
Tough Decisions
I don't come down on one side of this story vs. the other, but more of a slinky evolutionary path from one to the other. Some factors in my thinking:
Step 1
As it has for many of us in the next-gen space, innovation in session border control functionality is going to come from a variety of small companies, starting and competing in the space as they help define it.
In the short term, "SBC" companies will grow and proliferate, rather then consolidate and shrink, while the technology and the marketplace sorts out the value proposition, the customer requirements and the technology alternatives.
Enterprise network needs extend well beyond firewall traversal and NAT to things like CALEA enforcement, denial of service attack prevention, rogue RTP detection, etc. Enterprise business needs will, at least temporarily, support the viability of an expanding session border controller to serve their comprehensive data/packet/VoIP security needs.
Swerve. Step 2
Carriers of all stripes will require that same functionality, but on a network-wide basis, and furthermore, they will require it from either their traditional vendors or their next-gen softswitch/gateway vendors.
SBC functionality, is, after all, resident 1) in software and 2) in an IP network. At some basic level, therefore, it can 1) run on anything, 2) anywhere, at least from a functionality point of view. (Cost and control are another issue.)
Carrier networks will leverage these factors to demand consolidated functionality每more software in fewer boxes, from network equipment vendors每functionality at the network rather than the enterprise level.
Curve. Step X
Enterprises will have to decide if for issues of control they will require SBC functionality as CPE, or for issues of cost, they will be satisfied with shared functionality at the carrier level.
Current SBC vendors will have to decide if they're strong enough to be one of the very few surviving vendors aimed at the CPE market or whether they'd be happy to be subsumed into a larger network equipment function/company.
〞Mark F. Whittier, vice president of marketing, VocalData, Inc.
Economics Favor Softswitches
With the 2004 Presidential Campaign in high gear, a famous (not so) old campaign quote comes to mind when thinking about the challenge of handling session-oriented traffic (voice, video, etc.) through the IP network border: "It's the Economics, Stupid." While this is not quite what Bill Clinton's former campaign manager had in mind, the key idea remains. When exchanging session-oriented traffic across an IP network border, there are many ways of solving the interesting technical problems that arise but it is the economics that will ultimately dictate the solution that wins in the marketplace. Of the two solutions under discussion in this article (softswitches and session border controllers), which solution provides the lowest operational and capital costs for carriers? The answer, undoubtedly, is softswitches. The vast majority of traffic exchanged today is through the PSTN to IP networks. Carriers don't want to deploy and manage multiple boxes to exchange traffic with enterprises or other carriers just because the traffic arrives in a different format with different signaling. Softswitches (to be more exact, the Sonus Insignus Softswitch and GSX9000 Open Services Switch) can handle all types of traffic and all types of signaling. Session border controllers are confined to IP traffic. But it isn't enough to simply transport traffic. What is the lowest cost method of deploying complete services to calls arriving at the network border? Again, softswitches. Although session border controllers pass bearer traffic, they cannot play tones and announcements, transcode between various voice codecs nor offer services directly to traffic at the network border. They rely solely on multiple other costly devices to do that.
So, the question is why a carrier would want to spend scarce capital budgets and complicate the network infrastructure to deploy a completely new box that does not provide the complete set of services that carriers require.
They don't.
"It's the Economics, Stupid!"
〞Michael Rubin, director of product management, Sonus Networks.
SBCs Have Their Place
"Session Border Controller" is a catchall term that describes an IP network element that can provide a host of different functions. SBCs can be used as dynamic firewalls, proxy servers, edge routers, and lastly provide call routing functionality. It is not clear-cut that Soft Switches or Gatekeepers will assume all of these roles. In our view it depends on the size of the network and the applications desired. In smaller networks, it is very likely that a session manager will subsume all of these functions as this convergence makes sense and limits capital investment. In larger networks, standalone SBCs are the best solution available today since they allow for higher performance and scalability.
〞Chris Lengyel, product manager, Excel Switching Corporation.
Richarg Grigonis can be reached at rgrigonis@vonmag.com.
|
