Performance Consistency in the Cloud
summary
Cloud performance jitter affects P99/P99.9 more than averages and can cost millions—one case cuts tail latency by 100 microseconds and recovers revenue. Root causes: VM multi‑tenancy (L3 cache, shared thermal/power), managed EBS variance, and network placement. Concrete mitigations: use Nitro hypervisor and Nitro SSD controller, prefer local NVMe, run on dedicated hosts or full CPU sockets, pick instances with instance bandwidth configuration or fixed‑frequency ARM CPUs, test EBS types and enable SRD multi-path storage.
AI-written summary · about
chapters (11)
- 0:00Performance consistency intro
- 1:48Performance variation matters
- 3:57Latency variation problems
- 6:26Cloud abstraction unpredictability
- 8:12VM multi-tenancy jitter
- 9:55Mitigating cloud limits
- 12:07Shared resources impact
- 14:57Storage variability in block devices
- 17:50Reducing SSD tail latency
- 18:32Network path congestion
- 20:15Summary & Q&A
related talks
26mTuning Your Cloud: Improve Global Network Performance for Applications
42mStorage Performance Testing in the Cloud
16m"Performance Testing: ..."
24mBobtail: Avoiding Long Tails in the Cloud
34mOn hardware everyone is happy differently, in clouds everyone suffers the same. What to do?
48mMeasuring performance variabillity of EC2
32mBuild without limits with AWS Cloud infrastructure
26mTuning Your Cloud: Improve Global Network Performance for Application