Extending distributed tracing to kernel

Abstract

Quals talk:

Hundreds of services run on deep software stacks comprise large-scale distributed cloud applications. Debugging performance issues in such a system is increasingly difficult because different software layers abstraction, such as the application and kernel, limit observability across layers. As a result, the current debugging practice independently handles performance issues in different layers. However, engineers cannot correlate performance issues with root-cause from other software layers and cannot debug them successfully. In this talk, I will present my research extending the state-of-the-art debugging method in the user-space for distributed applications, distributed tracing, into kernel.

Join meeting in East Hall #016 or via Zoom.

Join Zoom meeting: https://tufts.zoom.us/j/3977553632

Password: see colloquium email

Dial-in not an option for this event.