Continuous delivery is supposed to help DevOps organizations quickly and smoothly deliver new functionality into production on an agile and incremental basis. In my experience application performance monitoring tooling has traditionally been thought of as a "last mile" addition or an after-the-fact project just as a deployment moves into production. This almost always means I will be spending my evenings and weekends digging into a production issue, whereas if we had put the application monitoring tooling in place at the beginning, I could be watching my favorite Star Wars movies instead!
I know this because I have lived this through pushing too many projects out the door without having the right visibility and living with the consequences. As a professional engineering collective we can actually do better by incorporating management telemetry throughout our delivery pipeline, bringing it in as early as possible, and seeing this as part of the application. This was our challenge at Oracle a few years ago when we set out to deliver a new cloud-based monitoring and analytics solution—built on a unified platform with machine learning and big data techniques against the full breadth of the operational data set—called Oracle Management Cloud.
With the latest March release of Oracle Management Cloud, developers can incorporate management telemetry early and continuously in the delivery pipeline builds, and readily benefit from the knowledge and discipline necessary to ensure high-performing releases and prevent production problems.
Developers like myself will really appreciate two things in this release:
1) The ability to maintain conceptual consistency across all of the environments by integrating continuous telemetry into the CI/CD flow, and
2) The ability to enable new code and features to arrive in production with a body of operational knowledge, like performance metrics, logs, and topological flow maps—already attached and ready to use.
As my own development and delivery organization has moved into the cloud, we’ve learned a lot from our experience. While there are numerous factors, there are three new areas in Oracle Management Cloud that I want to share in this series.
First Installment: Freshness of Monitoring
In this blog post I will dig into keeping the monitoring "fresh." From the artifacts checked into Artifactory to the application metrics, how do I know what I am looking at is representing the freshest and most relevant option?
Second Installment: Configuration as Software
I will then explain how I view the configuration of the monitoring and management software as part of the application, all the way down to source control and build scripts.
Third Installment: Automation
Monitoring needs to keep up with application changes and you need to make sure that the process does not require manual intervention. In this post I will go into the details of working through each part of the deployment process to make sure you are not incorporating a random monitoring and management tool that would become obsolete—but rather incorporate your solution as an application feature that grows/shrinks with the application and one that over time allows you to improve reliability and performance.
Join me for the first installment of the Oracle Management Cloud blog on "freshness of monitoring" on March 15th. If you are attending Oracle Code in San Francisco, be sure to attend the session "Gain Faster DevOps Insight with Continuous Monitoring and ML-based Analytics" (BRK1024-SFO) to learn more about the above topics.