
This video is currently available only for GOTO Chicago 2024
attendees.
Click to enter password.
Swimming Through Molasses: How Scale Makes Distributed Systems Slower
You need to be signed in to add a collection
Have you ever noticed that some systems don't seem to perform better as you scale them? No matter how many machines you throw at the problem, it never seems to get materially faster. Sometimes it feels like swimming through molasses. Have you ever wondered why? This talk will explore such phenomena by discussing how coordination and synchronization protocols introduce costly overhead. We will examine experimental evidence from both academic settings and industrial sources. Together, we will learn about mental and conceptual models to help structure problems such to prevent runtime slowdowns. Come for the distributed systems content, stay for the pragmatic solutions.
Transcript
Have you ever noticed that some systems don't seem to perform better as you scale them? No matter how many machines you throw at the problem, it never seems to get materially faster. Sometimes it feels like swimming through molasses. Have you ever wondered why?
This talk will explore such phenomena by discussing how coordination and synchronization protocols introduce costly overhead. We will examine experimental evidence from both academic settings and industrial sources. Together, we will learn about mental and conceptual models to help structure problems such to prevent runtime slowdowns.
Come for the distributed systems content, stay for the pragmatic solutions.