Ike Elliott, a friend and colleague, has a blog called Telecosm. It is worth reading.
The entry below was posted a month ago but it immediately sucked me into the fall of 1997. At the time, I worked for Kiewitt Diversified Group or KDG. So did Ike. You probably will not find any mention of KDG on our resumes as we changed the name a couple months later to Level 3 Communications. I always wondered whether the industry credited Ike for coining the term “Softswitch.” I was there when he came up with the term, so I know first-hand that he did. Ike shares a little-known fact in his post–Softswitch replaced the name “SIOS”, the SS7 to IP Operating System.
Thank you, Ike, for sharing this story. It inspired me to write down what happened in the two months prior to you joining us.
The Birth of The Softswitch by Ike Elliott
Posted January 09, 2008 on Telecosm
“O God! That men should put any enemy in their mouths to steal away their brains!” - William Shakespeare
When I joined Level 3 in 1997 to build the voice engineering team, I was immediately informed of The First Commandment of Level 3: Thou shalt not purchase any circuit switches.
Like Moses of old, my prophetic superiors had several memorable warnings to reinforce the rule: “The first person to sign a purchase order for a circuit switch is signing their pink slip,” and “Buying a circuit switch at a telecom company is like putting an open bottle of whiskey in front of and alcoholic.”
At the time, and for the century leading up to that time, circuit switches had formed the backbone of every voice network in the world. It was going to be real hard to build a voice network without them, but the challenge of building the first switchless voice network was what enticed me to join the company in the first place.
At least they had ideas about how to build this switchless architecture. They had a funky box-drawing on the back of a napkin, called the SS7-to-IP-Operating-System, or SIOS. I’m not completely sure, but I think is was Dan Caruso that came up with the SIOS name and did that first drawing.
My job was to take this outline and make it a reality, so the first thing I did was recruit a team of the finest engineers and program managers I have ever worked with, and likely will ever work with: Andrew Dugan, Steve Higgins (who hired Jin-Jen Wang), Carolyn James, Rick Steele, Bruce Baker (who hired Steve Branch), Ray Waibel. The list goes on and on. Some of the great ones were already there, like Rob Hagens, Kevin Dundon, Andrea Gavalas, and Jack Waters, and we got a few more later on, by acquisition, like Jon Peterson and, briefly, Shawn Lewis.
The next thing I did was coin a better term for SIOS. We were trying to emulate a circuit switch in software, so I renamed the thing the softswitch, and in my first presentation to Jim Crowe (who was still in Omaha at the time), Jim latched onto the name, and like an aggressive virus, the term started to infect the industry.
The softwitch name spread quickly because as we constantly talked with vendors about what we wanted to buy, the vendors started to use the term, too. Of course, none of them had a softswitch to sell us, so we were getting kind of worried about actually building a softswitch-based business. Several of us were commuting from Colorado Springs to Denver, and on the long car pool rides we would brainstorm about building a softswitch out of the then-limited industry “parts inventory.”
One of the options came from an unexpected source: a few weeks after I joined Level 3, Ron Vidal pulled me into Dundon’s office and described a company out in Cambridge that might have a softswitch. Ron and I paid a visit to XCOM Technologies, and liked what we saw, and later Jack Waters and I paid another visit, and on April Fool’s Day 1998 we announced the acquisition of XCOM for about $165M in overvalued Level 3 stock. We didn’t know it at the time, but that acquisition has to rank at the top or near the top of all of the acquisitions Level 3 has made.
XCOM had a rudimentary softswitch prototype going, built with Visual Basic, PCs, and some rubber bands, but the real important piece was that XCOM had developed a protocol for controlling dial-up-modem banks from their softswitch, and had sold Ascend on building that protocol into their network access servers. It was the first real application of the softswitch concept, even though it was initially designed for the “half-call model” of dial-up internet access. Shawn Lewis was the CTO of XCOM, and deserves a ton of credit for building the protocol first.
Despite all of the due diligence, we found that the prototype was not scalable and not ready for prime time. Even after the engineering team, under the direction of Hagens after Lewis quickly departed, did 8 months more software development and we launched the managed modem product, the softswitch was highly unstable and crashed at just about every meal time. At one point we were doing pre-emptive softswitch reboots every night, just to keep the memory leaks from causing prime time crashes.
It took more patience than the execs thought they had, and more sleepless nights than the engineering and operations teams thought they could sustain, to nurse the platform to stability. Our customers also deserve our everlasting gratitude for sticking with us while we worked out the bugs. But our patience and hard work paid off, and within 15 months of launching the platform, we had the most reliable managed modem service in the industry, with an 80% cost advantage over our competitors. Our market share quickly grew, and at its peak the managed modem business was responsible for over $800M a year in high-margin revenues for Level 3, before broadband internet access started to erode the dial-up subscriber base.
Any of us who were involved in the effort look back on those times with a mixture of pride and pain. We have the certain knowledge that we created a groundbreaking platform and a huge money-making business. However, the effort required a great deal of sacrifice from a great many people, which makes it very hard to repeat. Many other key people joined the team and made great sacrifices: Brent Bourne, Kevin Paul, Ken Fischer, Dan Rock, Jason Bach…far too many to list, but all of them crucial to the ongoing success of the business unit.
Some say that Level 3 could not have averted bankruptcy during the bursting of the Internet bubble without the softswitch. I don’t know if that’s true, because Level 3 had some pretty awesome financial engineers that frequently worked miracles of their own. I do know, though, that the softswitch made the job easier for those financial engineers than it otherwise would have been.
Why did we do it? The promise of valuable stock options played a role, yes, but I think it was a greater factor for the team to know they were doing important work, and to know they were part of a great team. And one thing I know, for sure, is that without the First Commandment of Level 3, we would not have been the first to build and deploy an operational softswitch.
And, at the time I left in 2003, we still had no circuit switches.
Posted by Dan Caruso (February 8, 2008)
