This course on Principled Performance Analytics aims to teach attendees an analytical method for analyzing and diagnosing distributed systems successfully. The course covers statistical methods used, the problem domain, and the intuition behind the approach. Attendees will gain an alternative lens for performance analysis and an understanding of potential pitfalls. The teaching method involves presenting the problem domain in detail, statistical methods, and real data examples. This course is intended for individuals interested in performance analytics, distributed systems, and reliability engineering.
Overview
Syllabus
Introduction
Is it working
Reliability
The Problem with SLOs
A Simple Example
How Does This Play Out
The Prize
Errors
Qualitative vs Quantitative
Performance Analytics Example
High Level Model
Two Sigma
Building Detection
FAQ
Real Data
Diagnosis
Excursions
Decoration
Taught by
USENIX
