Setting Performance Baselines for Java's 1-Billion-Row Challenge (Ep. 2) | With @caseymuratori

A video episode exploring how to establish performance baselines for Java's 1-Billion-Row Challenge. The hosts discover the challenge runs from a RAM disk rather than an SSD, which explains sub-second results that seemed physically impossible. They benchmark SSD read speeds using the `dd` command (~6.5 GB/s cold), compare Java file-reading APIs (BufferedInputStream, FileChannel, MappedByteBuffer), and find FileChannel matches the raw DD baseline at ~2.1 seconds for a 15 GB file. The discussion covers OS file caching, PCIe lane bandwidth, disk block sizes, kernel privilege transitions, and why chunk sizes matter for throughput. Homework for next episode: implement a working solution and set up a RAM disk for testing.