Lightning Talk - Benchmarking Persistent Data Volumes
summary
Discussion of why to benchmark storage and concrete methods: measure behavior under load, during replica rebuilds and failures, and after recovery; use dd for sequential tests and fio for configurable random I/O; use synthetic tests to exercise deduplication, encryption and compression algorithms and give fast developer feedback; run benchmarks consistently (commodity desktops are fine) and automate them into the CI build pipeline so a build fails on abnormal regressions; use application-level tests like Percona and Prometheus for database realism and test busy and failing systems.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:17Why Benchmark
- 1:00Synthetic Tests
- 1:50Developer Feedback
- 2:28Automating Tests
- 2:50Application Level Tests
- 3:31Summary
related talks
40mBenchmarking: Repeatable & Comparable
06m⚡ Lightning Talk: Back To Basics: How To Measure Etcd Performance And Not To Die Trying
40mPerformance Testing and Big Data
15mThe Case for Benchmarking Control Operations in Cloud Native Storage
10m⚡ Lightning Talk: Navigating the Complexities of Stateful Workloads in Kubernetes
38mScaling IO-bound application in the cloud
57mLast- und Performancetests in der Cloud?!
12mLightning Talk: Benchmarking Your Distributed ML Training on the K8s Platform