Interviewing Simon Eskildsen

Are our speakers awesome? Duh, everybody knows that! But what’s really under the surface? We’ve asked our speakers to answer a few questions about themselves so we can get to know a bit more about how their minds work.

Simon H. Eskildsen is a Production Engineering Lead at Shopify, where he protects the platform from flash sales, scale, misbehaving resources and itself. He’s coming to Full Stack Fest 2016 to talk about resiliency & DC Failover.

Seen anything great lately that inspired you?

Applying Pareto’s principle to everything. Pareto’s principle states that 80% of effects come from 20% of causes. When I started religiously applying this to decision-making and planning, it changed things for me.

Our company has 1,500 public Slack channels. I was in 20–30, listening in on many different teams’ discussions. I spent a decent amount of time doing this. Is this part of my 20% of activity that causes 80% of my output? No, so I left all but 5 of them. Does reading email right as they come in cause 80% of my output? No, the context switch is expensive, and often I don’t reply right away — now I delay all emails not directly sent to me to the next morning. Is spending dedicated time thinking about tasks that are important, but not yet urgent, part of my 80% output? Yes. Schedule 2 hours every week to think about something that may become a blocker months from now.

What is the last good book you’ve read?

Between the World and Me” is the best book I’ve read this year. It takes the form of a father writing a letter to his son about the struggles for him growing up in the poor neighbourhood of West Baltimore. This perspective on race, American history and diversity is an absolute must read. On page 20 I shed a tear. That never happens. Read this book.

High Output Management” is the best book on leadership I’ve read. The writing style is refreshingly human, and I’ve taken more notes on this book than any other. Doesn’t matter if you’re a people lead, technical lead or an individual contributor. This is a book about how to create the greatest possible results.

What’s the most exciting thing you’ve been working on lately?

I get obsessed with things in phases. In December, a tea store opened close to me. I went there to buy some tea. I asked a question, and quickly I discovered that this tea vendor spoke fluent Mandarin and Japanese to travel the tea gardens of the world to source the best possible tea. I went out dumbfounded and confused $100 poorer with a bag full of tea and “proper tea brewing equipment”.

Fueled with this momentum to learn, I devoured two books on tea and wrote 3000+ words of notes and created 100 flash cards on tea. Now tea is a bit on standby, with just tracking what I drink and how I like it in a spreadsheet. Now I’ve focused my attention on coffee. I’ve never been into drinking coffee, and I don’t plan to change that. I just want to get really good at brewing coffee. My feedback cycle will be purely through smell and other people, tracking what they say and teaching myself to incorporate input even more effectively in my cooking at home.

Can you share something about the current status of your work related to DC failovers?

Multi-dc we’ve already nailed. We have a script that fails over all of Shopify in a minute or two. Running in multiple datacenters concurrently, that’s hard and what I’m focusing all my attention on. It boils down to this: every unit of work must touch at most one shop, and every unit of work must be annotated with what shop it’s going to process. For example, a web request (unit of work) that hits `/internal/438998` is annotated with the shop id `438998` through the HTTP request. A job might have `shop_id = 438998` in its job parameters to indicate it will process that shop. This means you can have a layer in front of every unit of work that delegates it to the right shard for that shop.

For a SaaS application, this is all it takes. For an old application, that’s a tremendous amount of work — because it changes the assumptions 100s of people who’ve developed the application throughout the years have relied upon.

Does the near future satisfy your childhood expectation? How did you imagine it to be?

When I was 5, I wanted to be a magician. I thought, if you’re a magician you can just make money appear, and then you can do whatever you want. One of my clearest childhood memories was telling my dad that, and he told me “Well Simon, you need the money in the first place if you want it to appear.” I’m still not a magician, someone watching our team responding to incidents with terminals is probably the closest I’ll get.

Now the only thing I imagine in my future is that I continue to grow. That’s not a specific position. I don’t think my job will get more or less fun in the future, but it will have to continue to change for me to remain stimulating.


Simon is speaking at Full Stack Fest 2016 in Barcelona: Super early bird tickets are currently on sale. Don’t miss out the chance to see him & the rest of our amazing speakers on stage!

Buy your ticket now!

Note that we also have a Call For Papers open. We provide paid travel & accommodation for all our speakers, regardless of the selection process. Submit your talk and get early feedback so you increase your chances to get selected!


Simon on stage at dockercon14|eu

You can follow Simon H. Eskildsen on Twitter (@sirupsen) and on GitHub