Editorial : The Softswitch Is Dead: So What’s Next for VoIP Zealots?

Lucas

From the day in 1995 when Vocaltec demonstrated that you could use the Internet to deliver a voice call, the telecom equipment marketers and CEOs of start-up long distance providers have been trumpeting that voice over IP (VoIP) would replace the circuit-switched PSTN. That was hype, and it’s still hype. But before getting into why the softswitch is dead, first some background.

H.323

When Vocaltec and other vendors entered the VoIP scene in 1995, the start-ups needed a standard in order for their equipment to be plugged into the PSTN in Europe and elsewhere. The only standard close enough was the ITU-recommended H.323 for LAN-to-LAN videoconferencing via routers (or gateways) and IP networks. All that had to be done was to replace a video codec with a voice codec, and you were in business. H.323 included the concept of the gatekeeper for user authentication, authorization and usage accounting.

With a VoIP standard (H.323) and equipment, the then international call-back people and others saw an opportunity for international prepaid long distance. But H.323 products used low-bit-rate voice coders for efficiency and ISDN for call setup. One drawback in particular: there was no SS7 interface for carrier-grade networking. But it worked for prepaid calling cards, because the user just dialed via a traditional telephone as you would dial into a PBX.

Softswitch

Around 1998 the newly funded-long distance companies saw an opportunity to raise big bucks by saying that they had a switching concept that would revolutionize voice networking and make circuit switching obsolete. Circuit switches with TDM transmission would be replaced with a physically smaller router called a media gateway that would convert voice or video to IP packets for transmission. The call processing software that controls the distributed media gateways would be centrally located in a call agent or a softswitch, along with billing and OSSs. Also SS7 gateways would be included. In short, a new-generation voice carrier with a softswitch could interface with the PSTN like any other circuit switch. The benefits were that the interface devices (media gateways) were cheaper and required less electrical power and physical space.

The softswitch story sounded good, but a carrier customer was needed. The upstart IP-centric carriers were too busy flushing billions of investor dollars down the drain building fiber networks. They had no time to think of revenue, since it had no relevance to their stock valuation.

However, there was an ISP application generated by all the dial-up PSTN Internet users. ISPs leased primary-rate ISDN (PRI) lines from the ILECs to get dial-up customer traffic into their networks or remote access servers. A PRI is nothing more than a T1 circuit with call signaling via a separate 64Kbps channel call, or D channel. The problem—or opportunity, in the eyes of a softswitch vendor—was that big ISPs, with 100,000 or more subscribers, required many hundreds of telco PRIs that cost twice as much as a T1, even though they were equivalent in terms of transport. The rule is that ISPs are considered end users and not carriers, which are the only entities that can get T1s for intermachine trunks (IMTs).

So enter the CLECs. They could get IMTs from the telcos and save the ISPs millions. At the same time, they could collect interconnection fees from the ILECs. The only thing they needed was a network product to receive SS7 messages from the telco circuit switches and convert them into ISDN signaling (Q.931) for the ISPs. That product requirement was grabbed up by the softswitch people and dubbed the softswitch. The result: Lots of pent-up demand by ISPs and fast sales for the CLECs.

So, with a hot product (the SS7 to ISDN signal converter), and with PRI pricing inflated over equivalent IMTs, the softswitch vendors had a revenue stream and a story to promote to the investment community: “We have a product that’s going to replace every circuit switch in the world. And better yet, it rides the IP wave.” Of course every telecom, computer, and even the H.323 vendors began calling what they had at the time a softswitch.

Softswitch Reality: Death

First, the softswitch has seen so many definitions and product failures that the name itself is finished. But it failed as a PSTN replacement in the North American market for a number of reasons. Consider the following:

1) Not more efficient: The softswitch was hyped as being more efficient than circuit switching. It’s not. In order to be carrier-grade, the voice coders have to be the same as the PSTN (64 Kbps), you have to have expensive digital signal processing cards to handle fax traffic, and the IP network has to be segmented for voice-only calls to get an acceptable QoS. In short, other than saving space and electrical power, it’s not more efficient for voice than circuit switching.

2) No new services: The softswitch people say a kid with a Windows NT server in his garage could create new services overnight, whereas it takes years in the PSTN to create new services, because software development is controlled by the circuit switch vendors. Yes, this kid could program his service so an incoming call could have his home phone, a friend’s phone and cell phone all ring at the same time. If this kid could make three phones ring with one incoming call, a hacker could make a million phones ring with one incoming call. In short, the softswitch today doesn’t deliver anything new, and what could be new won’t see the light of day. Why replace a revenue-producing network (PSTN) with something new (softswitch) that in reality does less, not more?

3) Limited market: The softswitch market in North America is very limited. Prepaid international carriers are in bankruptcy or close to it. The PRI replacement market is shrinking as consumers go to DSL or cable modems for Internet access, and the startup, everything-over-IP carriers are near bankruptcy or already there. In addition, the CLEC interconnection fees from ILEC access are going away. Other than emerging countries with inadequate long-distance voice infrastructure, who’s going to buy a softswitch?

4) Class 5 misfit: Class 5 switches are local switches connecting your phone via a copper loop that gives you dial tone. They support analog phones and loops that are cost-prohibitive for softswitch support. They are underutilized, as well, with consumers dropping second lines in favor of DSL and cable modems. Support laws (E911, CALEA, or wiretapping and operator intervention) are written around circuit switch design and would need to be rewritten to accommodate softswitches. They also have billing systems, service provisioning and other OSS that won’t work with softswitches without massive investment, and on and on. Softswitches don’t have a place in the North American Class 5 world.

5) Open up competition: Even if the massive engineering economic problems stated above could be solved, for softswitches to succeed as a PSTN replacement, an ILEC would have to embrace them. But the ILECs would still reject the softswitch as a Class 5 replacement. There are a hundred and one reasons why softswitches would make it easier for competitors to go after ILEC customers. Here’s just one. Softswitches require less space in the central office (CO). CLECs, with their colocation desires, have in many cases been stopped cold because ILECs have no more space in their COs. With a softswitch replacement that barrier goes away.

So what is meant by “the softswitch is dead?” Anything that involves telephony and IP over the last three years has been labeled a softswitch. The name has no universally accepted meaning anymore. Certainly they won’t replace Class 5 switches, so they won’t replace the PSTN. What’s happening is that the basic building blocks—call control agents and media gateways—will find their way into VoIP H.323 products for enterprise networks, VoIP cable and 3G wireless. Also international carriers will send voice over managed IP networks for corporate users. But the name “softswitch” has no crisp meaning today. In fact the SIP people go out of their way to point out that their products are not softswitches.

Here Comes SIP

So if softswitches are dead, at least in the North American PSTN, what’s next? Session Initiation Protocol. This is a different approach to VoIP built around IP call setup commands (no DTMF signaling) and LAN-based phones (Ethernet vs. RJ-11 interfaces). The biggest service feature selling point is that you can simultaneously be connected to a voice and data connection (you can talk and at the same time send someone data or video). And of course you are plugged into the same LAN supporting IP data transport.

So is this the latest or next-generation VoIP concept for local and long distance, wireless and cable TV? Is it going to replace the PSTN as it is now being hyped to do? As with the softswitch, the answer is no. Is it a good topic for a future article? Yes. Meanwhile, if you need a better understanding of VoIP and the PSTN or 3G wireless networks, plan to attend TeleStrategies’ Understanding Telecommunications seminars. See www.telestrategies.com for a schedule of upcoming programs.