Josh Clark
Josh has both academic and real-world experience in the world of protocol analysis. He holds an M.S. degree in Computer Engineering with a focus in network engineering and has spent the past 8 years designing, troubleshooting, and optimizing networks and applications.
Sessions
This talk is an introduction to intuiting where non-network latency comes from. While it's usually quite clear how to determine what is network latency and what isn't, it's less clear how to dig into the timing differences between packets at different stages of a TCP conversation to direct troubleshooting at different layers of the stack.
Using a Linux-based web server as an example platform, this talk will demonstrate what network latency looks like, what host latency looks like, and what application/backend latency looks like. To explain what we see in the demonstration, we will also examine the web server to show how packets and requests propagate through the Linux OS to the web server application.
Attendees to this talk will leave with a greater understanding of how to identify latency at different stages of a web request. They will understand the basic Linux kernel and OS structure and how different stresses on a system show up in packet captures.
This talk explores how we might use Wireshark to optimize servers and applications even when they aren't slow. Depending on the type of traffic, optimizing TCP windowing and reducing the number of round trips required to transmit information can improve the speed of an application significantly.