Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn how to scale IP lookup operations to handle large databases using the innovative CRAM (CAM+RAM) lens framework in this 16-minute conference presentation from NSDI '25. Explore the challenges of designing scalable algorithms for modern networking chips like Intel Tofino, AMD Pensando, and Nvidia BlueField, which provide substantial ternary content-addressable memory (TCAM) and static random-access memory (SRAM) capabilities. Discover how the CRAM framework combines a formal model for evaluating algorithms on modern network processors with optimization idioms to address hardware-level constraints. Examine three new IP lookup schemes developed using this approach: RESAIL, BSIC, and MashUp, and understand their performance advantages over existing solutions. See how RESAIL enables Tofino-2 to scale to 2.25 million IPv4 prefixes compared to only 250k prefixes with pure TCAM approaches, and how BSIC achieves 390k IPv6 prefixes on Tofino-2, supporting 3.2 times more prefixes than pure TCAM implementations. Compare these results with existing state-of-the-art algorithms like SAIL for IPv4 and Hi-BST for IPv6 to understand the significant improvements in scalability for next-generation Internet infrastructure.
