How to lower the costs of your Drupal Site's resources and plan...
summary
Explains how capacity planning turns load tests and time series metrics into concrete buying decisions to avoid outages and wasted spend, using examples like a launch spike caused by thousands of outbound emails and a storage series at 23 TB nearing 30 TB that predicts running out in early April. Describes tools and methods actually used: fit YK and JMeter for curve fitting and load generation, StatsD/Telegraf for metrics, and ML forecasting from chapter 18 for autoscaling, with scripted workflows that proved predictive in demos.
AI-written summary · about
chapters (20)
- 0:00Intro
- 3:10User Expectations
- 3:47No silver bullet
- 4:47Capacity in Site Reliability Engineering (SRE)
- 5:14Resource management
- 6:34Quick and Dirty Math
- 7:38Demand Forecasting and Capacity Planning
- 8:08How SRE advocates for Capacity Planning
- 9:38Load testing a Drupal stack
- 10:06A Few Load testing Tools
- 12:14Identifying the key resources to measure
- 15:11Store and display time-series
- 15:24A couple of load testing tips
- 16:17Predict the future?
- 17:13Ceilings and Historical data
- 18:20Curve fitting
- 19:33Forecasting Peak-Driven Resource Usage
- 22:04Forecasting with Machine Learning
- 23:42Set a Goal!
- 25:01Capacity Planning Dashboard
related talks
35mCapacity Planning in Four Parts: Telling the Future without a Crystal Ball
51mLoad Testing: How to Know Your Limits
46mLoad Testing Without Being Off the Mark
26mLoad testing websites following the steps of a user journey
35mSurviving Black Friday: Tales from an e-Commerce Engineer
40mPerformance Testing and Big Data
34mLoad Testing Kubernetes: How to Optimize Your Cluster Resource Allocation in Production
57mLast- und Performancetests in der Cloud?!