Thanks. I think we should just give a quick round of hand to round of applause for Jason who started the microservices meet up because he saw a gap in New York and look at this.

It's awesome.

Okay, so what we wanted to do today is neither Adrian nor I wanted to prepare anything and thought instead what we do is have a great discussion around microservices.

We were talking earlier and thought what might be helpful is to talk a little bit about

just, I'm going to spend 10 minutes and talk about the approach we took at guilt.

And then Adrian is going to share one of his side projects.

He's actually still writing code in this language called Go.

I don't know how many people have heard of Go.

And it's really designed to help people understand and test some of the limits around microservices

and really learn how they behave.

It's actually pretty cool.

So we thought we do that first with the intention of really just seating some thoughts and questions

and then really what we hope for today is have a lively discussion.

Sound good?

All right.

So quickly, what is Gilt? Flash sales is e-commerce. We sell women's clothes, men's clothes, baby clothes, lots of things, events. If you're going out in the city, you should check out giltcity.com. Really cool things. We call it curated e-commerce. There's a lot of stuff out there, a lot of experiences, and our team at Gilt really selects the best, the top one or two percent in a given industry. So when you shop on Gilt, you know you're getting great things. I, as an engineer, have not changed how I dress, but I do buy all my clothes at Gilt. I like to say the only things that touch my body are actually my wife and clothes from Gilt. So there you go. One thing that's interesting when you talk about the people who are shopping on Gilt. Today, Gilt over half our revenue comes

from mobile. What I think is interesting is that about three quarters of all of our customers

interact with us cross-channel. So we have customers that only interact on mobile, but

the best customers today interact across all of the channels. And we really think about creating experiences for those.

So the beginning of Gilt, 2007, almost eight years ago, was simple. We started it on Rails.

This is what the stack looked like, Postgres. When things got hard, we just added some M-Cash-D,

and things worked. Right. And then this is what our traffic pattern looks like. So a Gil, it was interesting, is our new merchandise goes on sale at noon Eastern every single day. It's great products at a great price, and there's limited inventory, which meant a lot of people came at noon to shop. So in the early days of GILT, even in year one, we used to measure ourselves in Amazon, so we were half in Amazon a year in in terms of the traffic measured in request per second. And one of the things we learned is that acceleration, so the spike, if you measure the slope of the line at noon, that's a hard line. And caching actually doesn't help you because at 1159, the content is very different than what you need at noon. So those are kind of some of the technical challenges that we had to deal with. They were hard. This is how we recruit engineers of guilt, by the way. And like ThoughtWorks, we are hiring. So as the company was growing, we had some challenges, and challenges are euphemisms for outages. And so what we decided is that we needed to solve for that. So what we did is we really looked at the traffic funnel, starting at the top with login and registration, moving all the way down to checkout. And at the top, you have lots of traffic, and at the bottom, you have very little traffic. And we just start at the top and said, OK, so let's make login and registration fast. Let's build a micro service for users so they can log in and register.

...