If you had a motorcycle in LA and you were young, chances are you were like me, the last of the tuners just before the era of the superbike. Me, I had a Suzuki GS 550 and my mechanic at Three Start Sport Cycles on La Cienega hooked me up with the Yoshi pipe, clipon bars, and tweaked my cam. One year earlier I got the fever haning out with the Wednesday Night crew at the Crenshaw Burger King. It's one of those stories that got blended into the mythology via Hollywood, but the whole world and culture will never be known.
One of the legends of the era was a dude whose name I remember to be Glen. He lived around the corner from me on Somerset. His car was a matte black 1969 Camaro called Deadly Cancer. Obviously the man had worked on it for years, but rarely ever raced it. He said it was a 10 second car and I believed him - it was scary loud and he could smoke those dragstrip tires. It was faster than my motorcycle, and my motorcycle scared me.
One Wednesday Glen took the car out to Jefferson, down in what is now known as Playa Vista. This was, at the corner of Beethoven St, our drag strip proving ground. He killed his competitor on the first run. Second run, he threw a timing chain. Mighty Casey struck out. Fortunately, I had no money to bet with, having spent it all on stereo equipment. Glen never put Deadly Cancer back together. It was over.
I think about the prospects for garage software in the enterprise. Once upon a time, and periodically, I wished that I could work for three or four years on a single enterprise wide Hyperion project. I have to be the most proud of the handbuilt ETL I put together many years ago. And of course I'm immensely enjyoing my Ruby hacking. How would I feel if somebody came in and told me that my design was crap and I'm about to throw a timing chain? Not very friendly, probably.
This year I found myself over at SpaceX. I am enthralled by the possibility that we may start again in space. Elon Musk built the electric car that all of Detroit couldn't do, and basically didn't do. And now he's demonstrating that space exploration can be capitalized in the private sector where once it was only the province of governments, and he's working smarter. So once again, all things aerospace get my blood up. So in that spirit I watched a documentary on the life of the Concorde SST. Another thing we don't do any longer, proving once more quite clearly that stupid politics can triumph over brilliant engineering.
They said it was a 12 inch piece aluminum that fell off a creaky old DC-10 sitting on the runway that caused the crash of the Concorde in France that fateful day. But with airplane crashes, enterprise software systems and scratch built drag racers, it's never one thing. It's a combination of small things just slightly out of whack that creates, on that one rare fateful occasion, the perfect failure.
If only the block size were slightly smaller and the hit ratio on the index cache was retarded enough to free up cache memory, and that one hierarchy in the accounts dimension wasn't dynamic and that one user didn't query three years instead of two... Every binary 'if' doubles the possibilities but most systems design choices are not binary. There are five or six possibilities - we get weary counting them. So rationally we focus on the big things. What's the dense/sparse config? Are these tables denormalized? Is this process multithreaded?
But its the aggregate of small things too. We need to get the big things right or the small ones won't matter, but when the big things are right, we still need to perfect the small ones. They creep into the system like cancer...
NASCAR and F1, but mostly F1 is my gearhead fix. It is awesome to me to know that these small teams can change major parts of a race car in a matter of seconds. There's no way a thrown timing chain could end a racing career. That's because the team works together to win and they practice, practice, practice selflessly together so that everybody knows everything. That's how a system gets to live another day. Everybody knows everything. That's the difference between the proprietary garage work and the great power of open source. That's what is being proven in the field every day. It's the difference between survival and triumph.