December first made it a year here at Amazon Vancouver’s engineering castle in the sky. I’m working with good people in a cool office on interesting stuff. It’s at the white-hot center of server-side computing but surprisingly unsurprising.
Making vs talking · It turns out that building and shipping nontrivial software is a lot harder work than evangelizing it and writing about it. I come home awfully damn tired some days.
I’m less engaged in the Internet conversation and miss that some, but I’ve tasted more of that joy than almost anyone. And then the actual stuff I’m working on — making AWS more useful while keeping it reliable — is so blindingly obvious that it doesn’t really need evangelizing.
Yeah, computing is moving to a utility model. Yeah, you can do all sorts of things in a public cloud that are too hard or too expensive in your own computer room. Yeah, the public-cloud operators are going to provide way better uptime, security, and distribution than you can build yourself. And yeah, there was a Tuesday in last week.
Behind the scenes · Keeping all the world’s infrastructure on the air is a ton of work. Every day you have to balance risk reduction with shipping features. Bearing cloud growth rates in mind, 2018’s load could be 5× or 10× today’s. Thus we better be treating any current infrastructure creaks as Job Zero, and screw the shiny new.
But the high-tech biz has been all shiny-new all the time. Me, I think being reliable and available and fast in exchange for a monthly usage-based bill is the shiniest, new or not. Which is why I’ll never be a product manager.
What I do · I have a grandiose title and nobody can really tell me what to do, except for there are a million reasonable requests for help I really shouldn’t turn down, each an opportunity to feel bad about things fallen by the wayside. Fortunately I learned to ignore triage guilt a couple decades ago.
I get a few hours here and there to code, and thus a funny story: Sometimes my code-review requests came back with polite WTF’s about things that look like amateurish ignorance of obvious best practices. Finally someone asked why I was ostentatiously ignoring the common wisdom. So I shared the awful truth that in my lengthy career I had never previously written a single line of server-side Java. The silence that fell on the email thread was palpable.
But hey, at the end of the day, the server-side is all about message-exchange patterns and payload design and buying scale with sublinear algorithms, and I do know some stuff about that stuff.
So unless something goes terribly wrong in the next few weeks, some T.Bray code will soon be at work at scale, facilitating the Cloudification of online properties you spend time with.
How I feel · Corporately, I’m a burnout. I’ve co-founded two startups and worked for Sun and Google; most would see that as a trip through the Good Bits Of Capitalism. Still, I’m not a believer.
It’s stimulating to work for Amazon, which is approximately the most interesting company in the world by a factor of two. First, the pursuit of retail ubiquity; I won’t say at all costs, but with cheerful disregard for certain ratios thought important in the (detestable) finance biz. Second, the wholesale replacement of on-premise computing with Cloudstuff.
But at the end of the day, it’s a job, which is to say a financial transaction. My employer rewards me fairly, making a bet that what I help build will pay off for them. In exchange, I’ll work hard to help build things that do that.
But will I actually care? Enough to sacrifice family time or personal time or cottage-life time? Well, sometimes; when I’m convinced the work will touch people’s lives — especially my software-tribe peers’ — in a good way.
My whole career’s been blessed by luck. This feels like more. Don’t think I’m not grateful.
Anniversary resolution · Today, I tried switching my commute from bike-train-walk to bike-all-the-way. It’s refreshing, and just as quick; but we’ll see how my legs feel hold up.