Tom Herbert has 20+ years of industry experience at Intel, Facebook, Google, and Sun Microsystems driving development of networking solutions and operating systems at hyperscale
Relevant Links
LinkedIn
A preview of a new open source programming model and API for network protocol and other serial data processing!
Introducing Panda and opening words — 0:00
Tom’s background and experiences — 0:38
What’s the meta problem? Tom walks through — 1:43
Looking at the networking problem and the network processing bottleneck — 3:21
No one has solved this problem yet. Walking through existing models, processing types, goals, and shortcomings — 5:07
Moving towards a solution: write once, run anywhere, run well — 12:49
Gleaning some key ideas: domain-specific architectures, nature of the data, (fully programmable) CPUs are relevant, hardware acceleration, compiler optimizations — 15:20
Applying key ideas to solving the network processing bottleneck — 20:09
All of this brings us to PANDA: a programming model and API for applying these ideas to solve the network processing feedback —23:49
Sub-problem: Protocol parsing — 25:49
The PANDA Parser — 28:37
Details about the PANDA plan (to get announcements, follow Tom on LinkedIn — 31:23
*Share your questions and comments below!
Top comments (0)