With a focus on the larger implication of networking on performance, Artur will provide an exploration in why things are slow — and why it is our responsibility to understand our stack so we can fix it. Based on the experience of building and running a CDN with servers around the world that have one mission, deliver data as fast as possible, I’ll take a whirlwind tour around some of the issues:
TCP — how it affects us
New Zealand — you are very far away!
Monitoring performance in a distributed fashion
Reasoning about performance data and expectations
What it’s like to live in fear of cascading failure
Personal responsibility
And a little something about SSDs