Outage of the Sentry

Yes. Prezi recently suffered two site outages that prevented people from accessing their prezis for a total of almost four hours. It caused a lot of people serious problems, embarrassment, and stress. We’re sorry, and we owe everyone not just an apology but also an explanation. However, the pun is appropriate…

prox part 2 — akka streams with cats effect

Intro In the previous post, we have seen how prox applies advanced type level programming techniques to express executing external system processes. The input and output of these processes can be connected to streams. The current version of prox uses the fs2 library to describe these streams, and cats-effect as an…

prox part 3 — effect abstraction and ZIO

Intro The first post introduced the prox library and demonstrated the advanced type level programming techniques it uses. Then in the second part of this series, we experimented with replacing the streaming library from fs2 to Akka Streams. In both cases, the library used cats-effect for describing side effects. But it…

prox part 4 — simplified redesign

Intro In Part 1 I described how the advanced type level programming techniques can be used to describe the execution of system processes. It was both a good playground to experiment with these and the result has been proven useful as we started to use it in more and more production…

How Prezi switched data centers on the fly

The pit crew can change the tires on a racecar in 12–14 seconds[1]. That’s insanely fast compared to the 29 minutes, bruised fingers, and oil stained pants that it took me to do the same thing with our family car, but the difference is quantitative, not qualitative. Every single tire…

Effortless mass change management with Hammurabi

If you have ever developed microservices, you already know some of the challenges around it. If it’s done right, development can speed up, the cognitive complexity of the services can be decreased, and there are many many more advantages over monoliths. On the other hand, you have to solve different…

A Kubernetes crime story

The problem with having a great monitoring system is that you notice problems you couldn’t notice before. This is exactly what happened to us when we migrated our Nginx based reverse proxy to Kubernetes: we noticed that tail latency (p99) got somewhat higher for a few of our services. Increased p99 latencies start on 7/24 In…

Prometheus at Prezi: replacing 10 years of anti-patterns

What happens when you combine 10 years of architectural organic build-up and a lack of common best practices around monitoring? Bad on-call shifts, more outages, longer outages and a lack of technical flexibility. In our effort towards uniformity and reliability, we redesigned our entire monitoring stack and created a new…

Evolution of Data Ingestion and Product Instrumentation at Prezi

In this talk, we will cover how we took our need to better understand how our users use our product and how we ended up designing a system for product instrumentation and event processing to get those insights. Even though this does not sound hard, we were not as successful as we’d like to be at first. Thus we redesigned our data ingestion pipeline a couple of times to get to the state where we are today. We will start by covering how our data ingestion pipeline evolved, from starting with semi-structured event data copied to S3 with a bash script, to using Avro with Confluent schema registry ingesting events from Apache Kafka with Apache Gobblin to S3. Even though Apache Kafka helps a lot in scaling, just using Kafka alone is not your silver bullet. We had to introduce multiple components like using Avro format and the schema registry to solve for the missing pieces. We will also cover how we built around this ecosystem to make sure our system is foolproof; make it less painful to instrument the right events; and define how instrumentation works across the various platforms we support (Mac, Windows, iOS, Android, Web). Lastly, we will cover how this framework allows us to move faster, by benefiting data science projects and self-service analytics in general.

The Future of JavaScript

Startups identify with JavaScript. When you’re just starting out, you need to be dynamic. You need to be flexible. You need to be able to bust out a…

My Three Days at Craft 2014

Heartbleed Defeated

For those of…

Short-Lived Functional Teams Work — or — the QA SWAT Team

Functional teams have gotten a bad reputation in the startup community. There is no dearth of job listings for full stack engineers, or blog posts about the virtues of the DevOps movement. The consensus is that we shouldn’t partition expertise in a single team, and we shouldn’t restrict responsibility to…

For Our Daughters

Eszti The obvious answer is a shallow one: I have a two-year-old daughter and I want her to grow up in a world where…

Simpler Management of Modular App Development

This is…

In search of speed — debugging Elasticsearch performance

Background Here at Prezi, we use Elasticsearch to power the Explore pages for both Prezi Presentations and Prezi Videos. At the start of this year, we started running into performance issues with our cluster. Queries would take too long, some would timeout, resulting in 500 errors, high latency, degraded user experience…

Prometheus Unchained

A big shout-out is due to David Guerrero who did…

prox part 1 — type level programming in Scala

Intro I started writing prox at the end of 2017 for two reasons. First, I never liked any of the existing solutions for running external processes and capture their input/output streams. And I just returned from the scala.io conference full of inspiration; I wanted to try out some techniques and libraries…

Racing against race conditions

Some hours ago, we discovered a bug related to our Facebook/LinkedIn signup integration. It meant that in a very particular and unlikely set of circumstances, some people may have found themselves in someone else’s Prezi account. In the moments since discovering this problem we: mumbled expletives fixed the problem stayed…

Plans Are Useless, Automation Is Essential

Snakebasket — Recursively Install Python Dependencies

Performance Monitoring With Speedprobe

In the DevOps community, we all share common enemies: the insidious complexities of building and operating applications, the all-too reasonable, yet high expectation that users have for that application’s performance, and the scarcity of hours in the day. Time may be money, but it’s also our time, and the happiness…

Reactive Swift

UPDATE: ReactiveCocoa‘s new Swift API is out now. You can find my new post about how I upgraded this iOS app with the latest ReactiveCocoa on All The Flow blog. Reactive Swift: upgrading to ReactiveCocoa 3.0 If the title doesn't sound familiar to you, ReactiveCocoa is an FRP* inspired framework, counting remarkable usage in…blog.alltheflow.com Let’s go back in time to when Apple introduced Swift, the successor to good old Objective-C, it was a very…

Rendering: don’t try this at Chrome!

If, like me, you’re writing a graphics intensive web application that renders on Canvas, you may, with some justification, have thought that the era of cross-browser compatibility issues (and the inevitable accompanying hacks) was over. Well, here at Prezi, we’re working on a player to use Javascript and Canvas2D, and…

Technical Debt, a.k.a. “Not-quite-right-code” at Prezi

The…