CS Events
Computer Science Department ColloquiumProgramming Computer Networks Safely and Efficiently |
|
||
Monday, September 09, 2024, 10:30am - 12:00pm |
|||
Speaker: Srinivas Narayana
Bio
Srinivas Narayana is an Assistant Professor in the Department of Computer Science at Rutgers University. Srinivas's research goal is to enable developers to implement expressive networked applications with guarantees of safety and efficiency, by advancing domain-specific compiler and formal methods technology that improves Internet software and hardware. Srinivas received his M.A. and Ph.D. in Computer Science from Princeton University in 2016 and completed a post-doc at Massachusetts Institute of Technology in 2018. Srinivas's research has been recognized with distinguished paper awards at the Code Generation and Optimization (CGO) 2022 conference and the ACM SIGCOMM 2017 conference, and grants from the National Science Foundation, Meta, the Network Programming Institute, and the eBPF foundation.
Location : CoRE 301
:
Event Type: Computer Science Department Colloquium
Abstract: Over the last two decades, computing has been driven by a demand to process large datasets and build Internet services that support millions of users. While the demands on computing have exploded, the capabilities of general-purpose compute hardware have stagnated, due to the demise of Moore's law. Commodity network software stacks lag far behind the speed and flexibility desired in moving data between applications and processing data in transit.The prevailing approach to address network stack inefficiency is *software specialization*: moving away from the fully-featured network stacks in commodity operating system kernels, and avoiding most software code paths typically executed during data movement through the network stack. Specialization makes it possible to build high-speed applications, but it does not make it easy: Application developers are forced to also become experts in developing low-level system software.My research addresses three fundamental challenges standing in the way of effective software specialization for networking: safety, efficiency, and flexibility, by (i) developing principles andalgorithmic approaches to improve the efficiency of high-speed network software, with guarantees of safety and correctness; (ii) designing sound static analyzers to prove the safety of expressive low-level network software; and (iii) developing efficient compilation techniques to run expressive networking on high-speed network hardware.
:
Contact Professor Ulrich Kremer