To ensure safe and efficient deployment of Apache Kafka® clusters across multiple cloud providers, Confluent rolled out a large scale cluster management solution.

Rashmi Prabhu (Staff Software Engineer & Eng Manager, Fleet Management Platform, Confluent) and her team have been building the Fleet Management Platform for Confluent Cloud. In this episode, she delves into what Fleet Management is, and how the cluster management service streamlines Kafka operations in the cloud while providing a seamless developer experience. 

When it comes to performing operations at large scale on the cloud, manual processes work well if the scenario involves only a handful of clusters. However, as a business grows, a cloud footprint may potentially scale 10x, and will require upgrades to a significantly larger cluster fleet.d. Additionally, the process should be automated, in order to accelerate feature releases while ensuring safe and mature operations. 

Fleet Management lets you manage and automate software rollouts and relevant cloud operations within the Kafka ecosystem at scale—including cloud-native Kafka, ksqlDB, Kafka Connect, Schema Registry, and other cloud-native microservices. The automation service can consistently operate applications across multiple teams, and can also manage Kubernetes infrastructure at scale. The existing Fleet Management stack can successfully handle thousands of concurrent upgrades in the Confluent ecosystem.

When building out the Fleet Management Platform, Rashmi and the team kept these key considerations in mind: 

• Rollout Controls and DevX: Wide deployment and distribution of changes across the fleet of target assets; improved developer experience for ease of use, with rollout strategy support, deployment policies, a dynamic control workflow, and manual approval support on an as-needed basis. 
• Safety: Built-in features where security and safety of the fleet are the priority with access control, and audits on operations: There is active monitoring and paced rollouts, as well as automated pauses and resumes to reduce the time to react upon failure. There’s also an error threshold, and controls to allow a healthy balance of risk vs. pace. 
• Visibility: A close to real time, wide-angle view of the fleet state, along with insights into workflow progress, historical operations on the clusters, live notification on workflows, drift detection across assets, and so much more.

EPISODE LINKS

• Optimize Fleet Management
• Software Engineer - Fleet Management 
• Watch the video version of this podcast
• Kris Jenkins’ Twitter
• Streaming Audio Playlist 
• Join the Confluent Community

SEASON 2
Hosted by Tim Berglund, Adi Polak and Viktor Gamov
Produced and Edited by Noelle Gallagher, Peter Furia and Nurie Mohamed
Music by Coastal Kites
Artwork by Phil Vo

• 🎧 Subscribe to Confluent Developer wherever you listen to podcasts.
• ▶️ Subscribe on YouTube, and hit the 🔔 to catch new episodes.
• 👍 If you enjoyed this, please leave us a rating.
• 🎧 Confluent also has a podcast for tech leaders: "Life Is But A Stream" hosted by our friend, Joseph Morais.