by Amit (@amsharma) on Friday, 13 March 2015

0
Vote on this proposal
Status: Submitted
Technical level
Intermediate

Objective

This talk aims to discuss tools and techniques that can be used to root cause performance problems.
Key take away will be, a good know how of where to start looking when a performance issue crops up.

Description

Typical performance problems are described only in terms of their symptoms. Such as “my virtual machine is very slow”, “webserver is slow”, “the build is taking longer” etc.

During the talk I will discuss a few such problem situations and talk about the approach to root cause them, and possibly even fix them. Tools and techniques will be discussed as part of the approach.

Over the course of the talk. We will cover the following:

  • System observation tools for different system parameters (process, memory, io, cpu and network)
  • Profiling tools (ftrace, systemtap, dtrace etc.)

Requirements

  • Familiarity with linux

Speaker bio

Performance engineer at storage startup, Datagres Technologies (www.datagres.com).
Previously worked at Sun Microsystems, Cluster File Systems (lustre) and Oracle.