[$] The balance between features and performance in the block layer

Back in September, LWN reported on a seriesof block-layer optimizations that enabled a suitably equipped system tosustain 3.5 million I/O operations per second (IOPS). Thatoptimization work has continued since then, and those 3.5 million IOPSwould be a deeply disappointing result now. A recent disagreement over theaddition of a new feature has highlighted the potential cost of a heavilyoptimized block layer, though; when is a feature deemed important enough tooutweigh the drive for maximum performance?