Build - Test - Monitor: Microservice Monitoring for Developers on a CaaS Platform - DevOpsDays DC - 2017

09/01/17 • 21 min

For the past two years my team and I worked with a large federal agency to deploy & migrate to a new container-as-a-service platform based on Docker. The migration has enabled development teams to isolate components of their code for faster, more reliable development. But, we also saw that the additional tooling - such as monitoring technology - supporting these services doesn’t yet map to the model that developers need to efficiently monitor their own services. In essence, the Develop->Test->Monitor loop is still broken for modern environments. So how do you fix it?

This presentation is based on my real-world experience with container platforms. Based on this work, I’ll address: How do you effectively instrument your systems, without pushing too much burden on to developers? How do you isolate data, dashboards, and alerts in a way that improves security while simplifying analysis? What can you do to give developers deep information when troubleshooting, without giving them the keys to the kingdom? How do you facilitate data-driven conversations among your developers and ops teams?

If you attend this talk, you’ll walk away with tested, practical ideas that will help your teams become more self-sufficient, improve data-driven conversations among your teams, and evolve your monitoring infrastructure to work more effectively with your CaaS platform.

Previous Episode

Rolling Boulders Forever Uphill - DevOps with Docker, Kubernetes, AWS and Custom Slackbots - DevOpsDays DC - 2017

In Greek Mythology, the Gods cursed Sisyphus to spend eternity rolling a large boulder to the top of a mountain, where it would fall back of its own weight. In DevOps, we're forever rolling boulders uphill. We're making deploys faster, cheaper, smoother, and quicker. And once the boulder reaches the mountain top, the engineers rearchitect the application and the the process begins again.

At Upside Travel, Slack is our central command hub. We run our full operations through Slack ChatOps. Engineers request code reviews, product managers examine tickets, and the Slack-integrated NOC works slack-alerted events. We also manage our full continuous integration and deployment process through a custom Slackbot named, aptly, for the DevOps Greek hero, Sisyphus.

Sisyphus's simple promote command hides a complex dance of builds, tests, promotion, deployment and management. Upside combines Github, CircleCI, Artifactory, Terraform, Docker, Kubernetes and AWS to deploy code from nothing to something in 3 minutes and it takes deployment/promotion 100% away from DevOps and Engineering to place the power into the hands of Product Managers.

Next Episode

Lessons learned defending web applications when embracing DevOps

The standard approach for web application security over the last decade and beyond has focused heavily on slow gatekeeping controls like static analysis and dynamic scanning. However, these controls was originally designed in a world of Waterfall development and their heavy weight nature often cause more problems than they solve in today’s world of agile, DevOps, and CI/CD.

This talk will share practical lessons learned at Etsy on the most effective application security techniques in todays increasingly rapid world of application creation and delivery. Specifically, it will cover how to: 1) Adapt traditionally heavyweight controls like static analysis and dynamic scanning to lightweight efforts that work in modern development and deployment practices

2) Obtain visibility to enable, rather than hinder, development and DevOps teams ability to iterate quickly

3) Measure maturity of your organizations security efforts in an attack-driven defense model