Rootconf 2019

Observability for cloud-native workloads with Prometheus is the de-facto standard now. The next challenge in the space are high-availability, data de-duplication, data persistence and centrally monitoring multi-cluster workloads with the ability to do historical queries.

Thanos steps into this space and tries to address all by cleverly extending promethes’ features and using the power of cloud storage APIs.

Outline

• Prometheus Architecture

  • Lacunas
    • non-HA
    • non-persistent
    • Federation

• Thanos Architecture

  • Solutions
    • Extending prometheus with storeAPI
    • Deduplication in HA setup
    • Historical Queries on persistent stores

Speaker bio

Infra Engineer at Hotstar, previously of ZoomCar, Razorpay and CarIQ.
Kubernetes Geek. Dungeon Master and Boardgamer.