ConsumerTrack Onboards Kubernetes Apps in 3 Hours

About ConsumerTrack

ConsumerTrack™ helps brands connect with smart, aspirational audiences. Through our partnerships and owned-and-operated websites, we engage in-market consumers and connect them with the products and services that they are seeking.

Migrating to Kubernetes on AWS

ConsumerTrack were in the process of migrating to a microservices architecture running on AWS and Kubernetes. Approximately 24 services (4 existing & 20 new) would require new deployment pipelines and onboarding within a 12-month period. The Continuous Integration (CI) process (code to artifact) at ConsumerTrack was mature with tools like Git, Jenkins, CircleCI, Selenium, and BrowserStack. However, the Continuous Delivery (CD) process and deployment pipelines at ConsumerTrack still remained a manual ad-hoc scripted process owned by the DevOps team. “Onboarding a new Kubernetes service took on average 2 weeks for 2 DevOps engineers to manually script using Jenkins pipelines.” said Jeremy Malara, DevOps Manager at ConsumerTrack. The total effort/resource for onboarding these 24 new Kubernetes apps was ~96 weeks of effort and ~$290,000 in cost.

Manual, Scripted Deployments, Babysat by DevOps

Development teams at ConsumerTrack currently deploy 6-7 times per day.
“In the past, deployments were manual, ticketed and often babysat by 2 DevOps engineers and 1 developer from each dev team,” said Jeremy. Promoting code from each Dev to QA environment could take 2 hours (best case), with code promotion from QA to production taking 1 hour (best case). In addition, the verification/health checks of production deployments were manual, meaning DevOps and developers had to manually check metrics in Datadog/Prometheus and check application events in Splunk for a deployment to be deemed successful.

Deployments were manual and often babysat by 2 DevOps Engineers and a Developer - Jeremy Malara | DevOps Manager | ConsumerTrack

Self-Service Continuous Delivery for Developers

To automate and streamline their manual Continuous Delivery (CD) process, ConsumerTrack evaluated Harness. “In our Harness Evaluation alone, we reduced deployment time from Dev to QA from 2 hours to 5 minutes,” said Jeremy. The ultimate goal was to empower developers with self-service Continuous Delivery. Instead of DevOps engineers owning each deployment (and troubleshooting), they now govern and templatize the pipelines, which developers can instantiate using Harness pipeline templates and workflows from the DevOps team. However, the biggest win for ConsumerTrack and Harness was the onboarding of new Kubernetes apps. “Instead of 2 DevOps Engineers spending 96 weeks manually scripting Kubernetes Jenkins pipelines, we can do the same in Harness in one week,” said Jeremy. The effort and resource to create a new Kubernetes pipeline dropped from 2 weeks and 2 DevOps engineers to just 3 hours and 1 DevOps engineer. With Harness, Kubernetes pipelines are now automated end-to-end so code is promoted from dev to production in minutes vs. hours. Harness also uses unsupervised machine learning to automate deployment verifications using data from Datadog, Splunk and Prometheus.

Instead of 2 DevOps Engineers spending 96 weeks manually scripting K8 pipelines, we can do the same in Harness in one week. - Jeremy Malara | DevOps Manager | ConsumerTrack

First Year Savings of $580,000

Today, Developers at ConsumerTrack now deploy their own code.
ConsumerTrack also saw the following business benefits from Harness Continuous Delivery:

• Reduced deployment time (commit to production) by 92% from 3 hours to 15 mins.
• Reduced deployment effort by 16 hours per day equivalent of 2 DevOps FTEs ($290,000)
• Reduced onboarding time by 96% from 80 hours (2 weeks) to 3 hours per service.
• Reduced onboarding effort of 24 Kubernetes services from 96 weeks to 72 hours, the equivalent of 2 DevOps FTEs ($290,000).

The above benefits at ConsumerTrack translate to total year-one saving of approximately $580,000.