An implementation of the Jigsaw Sketch algorithm for finding top-k elephant flows in high-speed networks

The Jigsaw Sketch is a probabilistic data structure that combines the space efficiency of sketches with the accuracy of heavy-hitter algorithms. Key features:

• Space-efficient: Uses compressed counters and fingerprints
• Accurate: Provides accurate frequency estimates for heavy hitters
• Fast: Significantly faster than similar algorithms, e.g. HeavyKeeper
• Reversible: Can reconstruct original keys for heavy flows

The sketch maintains two types of cells:

1. Heavy cells: Store both frequency counts and partial key information
2. Light cells: Store only frequency counts for efficiency

For more details, see the original paper: "Jigsaw-Sketch: a fast and accurate algorithm for finding top-k elephant flows in high-speed networks" by Boyu ZHANG, He HUANG, Yu-E SUN, Yang DU & Dan WANG

• CMake (3.10 or higher)
• C++17 compatible compiler
• Linux/macOS

1. Create a build directory:

mkdir build
cd build

2. Setup CMake

cmake ..

3. Build

make

4. Download IPv4 data files

The project requires test data files for running examples and benchmarks. See the data/README.md for detailed instructions on downloading the test data files and information about the data format.

5. Run examples

Simple word_count example:

cd build
./examples/word_count ../data/war_and_peace.txt

IPv4 5-tuple example:

./examples/flow_processor

6. Run benchmarks:

cd build
./bench/sketch_bench