August 2000

• Defining The Softswitch
• The Burdens Of Success: Problems With VoIP

Defining The Softswitch

The June 2000 INTERNET TELEPHONY article by Jim Machi, "The Significance Of Softswitch" lays down a very broad definition of the softswitch. After reading this article I can see why. (While visiting SuperComm) I saw a lot of displayed equipment and reference to softswitch products. I was surprised by and amused by some of the references I saw. The term softswitch was being applied -- as Jim outlined -- very loosely.

In fact, the major telephone switch vendors, if they desired, could apply this definition to their legacy switches. When I think of a softswitch my mind's eye sees the three components that Jim described: Call Control, Broadband-Narrowband service integration, and Class 3/Class 4/Class 5 Switch functionality. But I also see the ability to allow third-party application service providers an interface to the softswitch. In this way, I as a service provider can introduce new customized applications for my customers. This is the real beauty of the softswitch.

So please include this third-party application programmability as one of the required capabilities of a softswitch and I am sure most of the loose use of the term will go away.
George C. Allen Jr.
Senior Member of Technical Staff
SBC Technologies Resources, Inc.

Jim has provided some very interesting insights about every piece of code now being a softswitch of some kind. I tend to agree, the industry needs some better definitions of gateway switches and softswitches. I think the press is going to have to do it, since the vendors seem intent on naming any lines of code a softswitch.

Alas, I don't agree with Jim's initial effort. He seems to be defining new Stored Program Control (SPC) host computers and applications that displace old SPC computers and the same applications as a good working definition of a softswitch. To me, replacing one centralized host application with another and calling it a softswitch is part of the confusion Jim is trying to clear up.

What I think he missed is a subtle client/server computing issue. Software code that substitutes one centralized Stored Program Control for the other or simply puts a centralized Stored Program Control application on certain traffic in a PBX, Class 4 or Class 5 configuration are SPCs and operate more like gateways and gatekeepers than softswitches.

From my perspective, softswitches are client server computing environments, with the client being the switches and gateways that use a server to control the establishment of services, via shared resources that may or may not be in the client device. These resources can be attached via a procedural call issued by the server as a proxy for the client switch or gateway. The Server application may or may not take over the direct control of a switch (as used here I define switch as the SPC application that connects an input port to an output port).

What the softswitch server does for the client switch or gateway is to take over the role of determining and implementing the network resources required to satisfy the client request. Once these resources are identified, the server instructs the gateways and switches on what to do to enable the shared resources to be deployed at the quality and policy level of the service supplier. This proxy functionality is implemented independent of the stand-alone centralized gateways and switch applications. This client server approach reduces the switch and gateway to a thin client device that can be optimized for low cost connection of inputs to outputs.

In my mind, devices that are not working in this client server environment are typically called media gateways. The gateways convert analog and digital signals to optical or radio transport signals for transport over networks. For example, a Class 5 by definition is a media gateway to a SONET or asynchronous transport network. Putting all that Class 5 code in a PC doesn't change the application nor make it more efficient or effective.

Today, I think of a softswitch as part of the new distributed processing applications where servers identify and commit the shared network resources to client devices that request them -- independent of the protocol they use to make the request (SS7, TCAP, MGCP, SIP, touch tone, rotary dial, multi frequency, A and B bits are just names for protocols to move information across a media...).

Of course my client/server definition may only lead to computing guys confusing the softswitch issue with different terminology. On the other hand, a call server and a connection control server seems easier for me to understand than today's any code that connects two ports is a softswitch.
- Howard J. Gunn
Vice President Marketing And Sales
TeraBridge Technologies Corporation

Jim Machi responds:
One of the purposes of the Industry Insight column is to get readers thinking about issues that potentially have not been brought to the forefront yet. While the "Significance of Softswitch" column discussed the softswitch in what I referred to as a distributed architecture environment, Howard correctly realized that this indeed is a client/server environment.

The Burdens Of Success

Well, you can say what you want about the "success" of VoIP, but the three times I've tried to use Net2Phone, the best I can say is "what a piece of garbage!" I was never able to even test it. It appears that the site was so overloaded you couldn't get anything out of it.

I hope there are "real" success stories out there somewhere. And no, I do not work for any circuit-switched company. I'm simply a software developer who has done some telephony work.
- Brian Keller

Marc Robins responds:
First, thanks for your message. I always appreciate when a reader takes the time and energy to respond to something I've written.

With respect to the problems you've encountered with Net2Phone, I must admit I've heard similar tales recently. According to managers at Net2Phone, they are enjoying too much "success" and are struggling to accommodate the surge in call requests. Keep in mind similar problems hit leading e-commerce companies during the last holiday season, when they didn't add enough capacity in time to accommodate the surge of online shoppers. And remember when AOL was generating more busy signals (and bad PR) than profits?

Now, does a lack of adequate capacity in the face of surging demand equate to a wholesale failure of the technology and business model? I, for one, think not. You say that due to an overload situation, you couldn't test the service. That is definitely frustrating and a big turn off. But if you couldn't actually use the service, how can you say it's "a piece of garbage?"

I hope that you will give the technology another chance. And if you've given up on Net2Phone, try some other Web-based phone services, such as www.dialpad.com, www.hottelephone.com, or www.phonefree.com. And let me know how it goes -- I'm really interested in your experience. Just remember, with any new groundbreaking technology and service, there are bound to be bumps in the road to a problem-free customer experience.

We invite readers to send their input and advice to ggalitzine@tmcnet.com.

[ return to the August 2000 table of contents ]