@kodekloud: System Design: Metric Collection Pipeline for Microservices Prometheus is great at scraping metrics, but it stores everything on one pod's disk with one retention setting. Once data ages out, it's gone. In this short, we break down how Thanos decouples compute from storage, offloading sealed metric blocks to Amazon S3 and stitching fresh and historical data back together so Grafana sees one seamless source of truth. How long do you retain your metrics today? Tell us in the comments. #prometheus #thanos #kubernetes #systemdesign #devops #observability #grafana #aws #s3 #cloudnative #sre #monitoring #microservices #platformengineering #infrastructure #devopsengineer #cloudcomputing #k8s #techshorts #kodekloud
Great design and well explained! I have built similar designs in Azure.
2026-06-11 20:18:14
0
hr.1233 :
Can we have a dedicated course on monitoring
2026-06-11 20:42:26
0
Eric Orr :
I feel like this is over engineering what is essentially a message queue. Like you have a bunch pubs going to two redundant subs, then you build a queryable structure from that, but I feel like RabbitMQ and Dynamo DB achieve this in a simpler fashion without dependencies on Prometheus and Thanos.
Correct me if I'm wrong, I'm only a student.
2026-06-12 08:28:04
0
amged ahmed :
cool
2026-06-11 16:36:09
0
To see more videos from user @kodekloud, please go to the Tikwm
homepage.