Silicon Valley 90’s Style

How does a kernel-hacking Georgia boy wind up in the middle of Silicon Valley?

Posted on by

(This week’s amazing picture is from Anthonyavalos408. CC BY-SA 4.0.)

It was my first ever California interview, and it seemed to be going well. I sat in a stuffy room that contained a chair, a sofa, and a studio piano. Turns out, a star engineer also played beautifully, so the company covered the walls of an office with speckled acoustic tiles for soundproofing and called it a day.

This particular soundproof office was located in a Redwood City complex at the northern limits of Silicon Valley, the same office park that housed Steve Jobs’s company Next. The place really did look like a park. The low, single-story buildings were surrounded by green grass and lush trees.

It was early on in 1990, but so far that year I’d met with engineers, sysadmins, and several folks from top-tier institutes of higher learning all of who had various letters listed after their names.

Then in walked Audrey MacLean, CEO.

She sat down, greeting me warmly with the same high-energy manner she has today, and said “I’m not here to see if you’re the right for Adaptive--I’m here to convince you to come work for Adaptive.”

I’d first met Audrey about a year earlier. She and Charlie Giancarlo were on the hunt for a telecom technology for a new spin out they were working on. I remember their visit to our lab and the follow-up visit by one of their top programmers.

At the time I was working in Norcross, Georgia, for Digital Transmission Systems (DTS) building VME 68020 based systems to control large telecommunications. Charlie had done some preliminary work with DTS before my time there and had since headed west to California. During a whiteboard session with Audrey, he had a great idea for a new company and DTS came to mind. Soon the two were walking through the lab.

Control systems were the brain for DTS’s equipments. It wasn’t just a matter of using a timesharing system and clicking some other things together, but the decision to use Unix did make things a lot more possible and simple... if the designer had the freedom needed.

At DTS, and later at Audrey’s company Adaptive, I had those resources. For example, there needed to be some sort of network for telecom system equipment. Some shelves were interfaces to the long distance wires, called “facilities,” and others were switches that pulled the different channels apart and put them back together again. Everything had to be logged, alarms needed to be sounded, relay closures closed and opened, and cutoff buttons read.

A fully configured DTS switch was six 24” racks of equipment. We used a variant of IBM’s High-Level Data Link Control (HDLC) protocol running over RS-485 multidrop cables. We had some good silicon for HDLC to use both in my controller and in all the small eight-bit processors in all the shelves.

Some of my first protocol designs were getting all the parts to talk and it worked out really well. Most of the shelves must appear to higher level software as systems with a number of special purpose registers. My protocol just fetched and set those registers. The firmware in the controller would take action when certain values were stored in certain registers.

A few days after Audrey and Charlie left, Milan Momirov arrived. I remember the timeline of these visits vividly because it was in late 1989 as Betsy, my wife, was in the final weeks of pregnancy with our first little girl.

I was told that Milan was on a deal-making due diligence mission for Audrey and Charlie. Later I would learn that they wanted to buy DTS for our switching technology.

Milan asked to see the software. This wasn’t an issue because at the time I kept a pretty up-to-date bound listing to reference. I made changes in pen as I went. If I made significant additions, I’d just print a new copy. So I handed him that.

A short while later, he asked to see me. We sat at an empty desk in one of DTS’s windowless Norcross offices, my listing laid out in front of us. Milan began to ask me questions about the system.

“Where does it protect a failed card?”

I flipped around in the listing, answering, “We do that right here.”

“What about the finding a replacement link?”

“We did that in this code,” I responded, turning to another page.

And so it went.

Although I don’t remember the details of our conversation, I do remember enjoying it. Milan was sharp and caught on to what I was doing quickly. It was a great opportunity for the kind of in-depth conversation that only occasionally happened at DTS. There were many good folks there, but Milan was very good.

He left the next day and I forgot all about. And then my phone rang.

“Can you come to the conference room?” asked our CEO’s assistant. As I walked to the front I was wondering what I’d done. Maybe they’d grown tired of me arriving at noon every day. Most people didn’t see me leave in the wee hours of the morning for my 70 minute commute home to the woods. The radio was usually beginning to broadcast Morning Edition as my day ended.

To my surprise, I was ushered into a conference room where Charlie Giancarlo was waiting. Alone. We sat down, exchanged a bit of small talk, then he pulled out the copy of the source code I’d given Milan.

“When Milan was here,” he began, “whenever he asked about the code, you always answered with ‘we’ did this and ‘we’ did that. I came here to find out exactly who is ‘we.”’

I’ve always said “we” a lot because it’s how I was raised. My father started many businesses and I quickly learned that it’s best to speak of the company as “we” and not “I.” It’s never about “I,” even when there’s no one else around.

“I grew up in small business,” I said. “‘We’ reflects best on the company. In this case, we is me.”

“Well, ‘we’ at Adaptive thought so.” And with that, he left.

Adaptive didn’t buy DTS. There was a fox in the hen house that wanted us: the Irma board people who were just over in Alpharetta, DCA, a sworn enemy to Adaptive’s parent NET. This sparked a bidding war, and soon Audrey was inflicting maximum pain by driving up the price so DCA would have to pay dearly for us. Audrey went to Tell Labs, buying their technology for Adaptive’s systems.

My project was put on a maintenance plan probably because DCA didn’t know what to do with us. It only took one depressing meeting with the new manager to know that I needed to leave.

What about a call to Adaptive? What could it hurt?

So I rang Charlie’s assistant. “This is Brantley Coile, can I speak to Charlie Giancarlo?”

The assistant later told me that my name sounded pretentious and apologized for being rude to me. I was honestly able to say she hadn’t been rude, and let her know that when I feel pretentious I go by B. Webster Coile, the 3rd.

Not long after that phone call, an envelope arrived with a plane ticket. The arrival was late at SFO. It was the first time I had to find the rental car place, get down 101 to Foster City, check into the Courtyard by Marriott, and sleep. The next day, it was down 101 to Seaport Boulevard and into the small piano room.

So, there I was, dumbfounded that I was sitting in front of this bright-eyed, high energy CEO who just declared she wanted me to come to California. This is not usually how my interviews typically went.

Usually, I spent most of the interview trying to figure out how much the interviewer knew, so I could best explain what I did. Sometimes it was great. Mostly it wasn’t. For example, interviewers respond badly when you point out that there is a fundamental logical flaw to their question.

Audrey continued the conversation by saying, “When Milan came back, all he could talk about was Brantley, Brantley, Brantley...”

The next day, I met with the VP of engineering, Roger Chung, this time in the little sandwich shop in the office park (the same shop where I first saw Steve Jobs). He spoke softly, so softly in fact, that I didn’t realize there was a signing bonus to the job offer until I got the check. It was a really nice surprise.

I was soon all set up in the Adaptive family. The earthquake of ’89 was still being cleaned up and large trucks with debris from the Embarcadero would chug past our complex, headed to the hammer mills at the end of our causeway. We were right on the bay, and barges would take the ground-up concrete for recycling. There was a car recycler down there as well, shredding up old cars, putting them on barges towed up north before their trip to Japan only to return to the U.S. as Hondas and Toyotas.

My first order of business was a meeting in Berkeley with a great company called Mt Xinu. They were founded by a bunch of CS majors who wanted to make a living working on the Unix kernel and did really neat things. That day we met with someone, whose name escapes me, that Adaptive approached to do a Unix kernel for our new telecom system. Bob Kriddle started off the meeting. I got the feeling he was the business brains.

I listened to their plan to port the Mach kernel to our Motorola 68030 board and put Unix System V on it, and how we could use Mach’s fancy message passing features in our system.

As we drove over the Bay Bridge something wasn’t quite right. System V was big enough. Sitting on top of Mach was absolutely huge. I had my doubts.

Back in my cubicle, I got out the Mach docs, looking for something specific. I was looking for the equivalent of BSD’s “An Introductory 4.4BSD Interprocess Communication Tutorial” document. This cookbook would help figure out how to use the system calls to do messaging, how one program sent messages to another in the same system. Fifteen lines of code to open a socket. Another half page and there was a primitive server. Another few lines and a client could connect to the server. It was all very simple.

After a while, I found what I was looking for. The simplest example of using ports and messages was spread over ELEVEN pages!

At that point, be it formed by nature or nurture, my simplicity gene sounded the alarm. No way. Nothing doin’. Bad idea. Run for the hills. This dog don’t hunt!

I stormed out of my cubicle with the doc in hand, headed down the crowded hall to Roger’s only marginally larger cube, and said very much the above. On his wall was a transparency that said “Simple Solution Wins.” I pointed at it and said, “This ain’t it!”

“Okay, what’re you going to do about it?” What I did is a story for next time.

Part 2 →

Software Efficiency Puts Cash in Your Pocket

I’ve learned a lot about storage over my career. Even in the 90s, I could tell you that well-written, efficient code saves countless dollars. You, too, can save cash on software with Coraid EtherDrive storage appliances.

Save Money on Data Storage

If you have a moment, use the form to drop me a note about your storage needs so I can help you save money using fast and high-performing Coraid storage—less than $0.001 per GB a month!

About the Author

Brantley CoileInventor, coder, and entrepreneur, Brantley Coile invented Stateful packet inspection, network address translation, and Web load balancing used in the Cisco LocalDirector. He went on to create the Coraid line of storage appliances, a product he continues to improve today.

Sign up to have interesting musings delivered direct to your inbox.