Building a K8s Operator for a Distributed Database

on YouTube ↗21m

" summary

Kubernetes operator that keeps Aerospike clusters online through live patches, replacing live clusters of tens of nodes and recovering crashed nodes to meet a claimed 100% uptime. Implements an init‑container wipe flag to distinguish new volumes from restarted pods so persistent data is not lost, and a reconcile loop that requeues after each operation (with API response delay) to allow immediate aborts of rolling updates. Notes scale limits: customers with ~half a trillion objects require private hardware and overprovisioning beyond what an operator alone can solve.

AI-written summary · about

" chapters (6)

1. 0:00Introduction
2. 1:23What is a Kubernetes Operator
3. 7:20Aerospike
4. 11:24Persistent Data
5. 13:23Rolling Updates
6. 16:34Large Scale

" related talks

• 16mBuilding a K8s Operator for a Distributed DatabaseNatalie Pistunovich · Conf42 Cloud Native 2021
• 35mOrchestrating Resilient Data: Harnessing the Strength of Kubernetes with OperatorsGregor Bauer · stackconf 2024
• 44mDatabases on Kubernetes: Why You Should Care!Denis Wilson Souza Rosa · GOTO Oslo 2020
• 29mFlipkart: Scaling Aerospike at Flipkart with KubernetesSharath B. P., Sandesh Kumar Gupta · Aerospike APJ Summit 2023
• 15mHow We Solved the K8s Cluster Communication ChallengeDavid Gerchikov · KubeDay Israel 2023
• 41mDatabases on Kubernetes Using a Custom Operator: Day 1, Day 2, and BeyondJohannes Unterstein · KubeCon + CloudNativeCon Europe 2019
• 21mStory of a Kubernetes Go OperatorAravind Putrevu · Kubernetes Forum Delhi 2020
• 24mWhat a Modern Database EnablesSrini Srinivasan · Aerospike Summit 2023