If you follow a typical playbook for digital transformation in the enterprise it will look something like this:
Migrate from desktop apps to mobile apps
Migrate from data centers to public/hybrid clouds
Migrate from monolith applications to microservices architectures
Migrate from waterfall/agile to continuous delivery
Back in 2014 AppDynamics announced support for native mobile apps and has also had several major announcements around support for AWS, Azure and most recently Pivotal Cloud Foundry.
A few weeks ago we introduced our Kubernetes support which was a significant step to supporting microservices architectures. Today, we’re announcing a new partnership with Harness to help our customers embrace Continuous Delivery.
What is Continuous Delivery?
Continuous Delivery is the ability to get changes of all types—including new features, configuration changes, bug fixes and experiments—into production, or into the hands of users, safely and quickly in a sustainable way.
Who is Harness?
Harness offers Continuous Delivery as-a-Service that allows customers to automate how software is deployed and delivered to end users in production. Harness was recently recognized as a 2018 Cool DevOps vendor by Gartner.
Why did AppDynamics Partner with Harness?
Unlike traditional Application Release Automation (ARA) solutions, Harness takes Continuous Delivery one step further by verifying deployments using APM performance and log data. If performance anomalies or regressions are found post-deployment, Harness can automatically rollback the application to the previous working version.
Harness allows our customers to move fast and not break things using insight from AppDynamics.
More importantly it helps our customers answer the most important question:
“What is the business impact of every production deployment?”
Understanding the Business Impact of Production Deployments
Using AppDynamics APM and Business iQ customers can monitor the performance and revenue of their business processes/transactions for each application and environment.
Harness is able to query this business and performance data within seconds of a new deployment, and use machine learning to build models of what normal performance and revenue looks like.
For example, see the below deployment workflow in Harness which illustrates the deployment of a new application service into production.
You can see once the service was deployed that Harness automatically connected to AppDynamics to perform a verification (aka health check) which failed.
If we click on the red AppDynamics verification icon we see why this specific health check failed:
Harness machine learning detected three high risk transactions, specifically:
2 Performance regressions for the Register and Login transactions
1 Revenue regression for the Checkout transaction
By mousing over each regression you can see the values behind the machine learning verification. In the case of the Checkout transaction, revenue per minute dropped from ~$14k/min to $3,200/min after the new version of the application was deployed.
As a result, Harness automatically initiated a rollback to the previous working version:
Think of this as a safety net when verifications or health checks fail.
Over the coming months we’re looking to deepen our integration with Harness. Stay tuned for more on that!