What is CQRS?

Bertrand Meyer first introduces the principle of Command Query Separation in his book Object-Oriented Software Construction. The principle states that a well designed object should have methods that are either commands or queries. A command changes the state of an object, but does not return any data, while a query returns data and does not change any state. By dividing methods into these two categories, you will have a better understanding of what does, and what does not, change the state of your system. [Read More]

Dissecting SQS FIFO Queues — Does Ordered and Exactly Once Messaging Really Exist?

At first glance, Amazon’s First-In-First-Out (FIFO) message queues provide an excellent feature set for business-critical scenarios. With FIFO, the order in which messages are sent and received is strictly preserved. With exactly-once processing, message duplicates are not introduced into the queue, and consumers control when a message is made available for redelivery. Reading past the marketing hype, how well do FIFO queues work in the real world? This article takes a deep dive into SQS FIFO queues to test the claims of message ordering and exactly-once processing, paying particular care to the conditions under which these claims hold. [Read More]

Learning a Language with Amazon Polly and a Serverless Chalice App

For the past year I’ve been making a concerted effort to learn French using the methods from the book Fluent Forever, which is an excellent resource for learning how to learn a language. For those not familiar with the method, it boils down to this: Learn Pronunciation: knowing how to correctly pronounce words in your target language makes everything else easier. Learn Frequently Used Words: not all words are created equal, learn the most frequently used words first. [Read More]

Evolving Messaging For Microservices: A Retrospective from Building Workiva’s Messaging Platform

Workiva’s original product — supporting the mundane task of filing documents with the SEC — was so innovative that within its first 5 years it was being used by more than 65 percent of the Fortune 500 and generating more than $100 million in annual revenue. During that explosion in growth, the software development team focused solely on supporting and expanding the existing software stack. However, after several years of growth and expansion maintaining and extending that single code base became unsustainable. [Read More]

Ordered Messaging in an Unordered World

At Workiva, we’ve been using NATS to provide a highly-available, scalable message delivery service. The tradeoff for having these properties is that there are restrictions on the order in which messages are received by subscribers. NATS delivers messages from a single publisher in the order in which they were published. However, once you introduce multiple publishers, no such guarantees exist. Furthermore, even if messages are successfully delivered, they may not be successfully processed — if a subscriber fails to process a message correctly when it was sent, it may become out-of-order after retrying. [Read More]

How to create a functional VPC using CloudFormation

This tutorial walks through how to create a fully functional Virtual Private Cloud in AWS using CloudFormation. At the end of the tutorial, you will have a reproducible way to create a virtual cloud with three subnets, a security group, and an internet gateway with SSH access for your IP address. I’ve found this template useful for creating an isolated environment to develop and test software. Full code for this tutorial is available on Github. [Read More]

Compiling private Java code from Leiningen

For a recent project, I wanted to verify the correctness of a distributed queue implementation based on Amazon SQS. For this, I turned to the Jepsen library for verifying distributed systems. Jepsen is written in Clojure and the first task was to get Jepsen to compile with a Java library hosted on our internal Maven repository. I googled for a while, asked around, and assembled instructions from a few different places. Here then, is a single blog post summarizing the solution for future use.

[Read More]

Uploading Large Payloads through API Gateway

API Gateway supports a reasonable payload size limit of 10MB. One way to work within this limit, but still offer a means of importing large datasets to your backend, is to allow uploads through S3. This article shows how to use AWS Lambda to expose an S3 signed URL in response to an API Gateway request. Effectively, this allows you to expose a mechanism allowing users to securely upload data directly to S3, triggered by the API Gateway. [Read More]

Getting Started With TLA+

This post shows how to write your first simple TLA+ specification. What is a specification? In software, the behaviour of a system is described as a sequence of states. Mathematically, each state is expressed as a function F(t), which represents the state of a system at time t. To completely specify a system, we write out each state to fully define the systems behaviour. A simple clock This example comes from Chapter 2 of the book Specifying Systems by the creator of TLA+, Leslie Lamport. [Read More]

Basic Math for TLA+

At its most basic, TLA+ is a written description of what a system is supposed to do. More specifically, TLA+ is a specification language for formally defining the behavioural properties of a system. TLA+ is based on temporal logic, which is built on top of first-order logic and set theory, and provides some conveniences for working with large specifications for complex systems. You can think of a TLA+ specification as mostly ordinary math and logic, glued together with temporal logic for parts requiring it. [Read More]