4 ms·
Even for 2006, what would make this argument for me is actual numbers. Is it plausible that some programs worked well like that then? Sure. But I can promise no
by wpietri 6y ago
Even for 2006, what would make this argument for me is actual numbers. Is it plausible that some programs worked well like that then? Sure. But I can promise not everything did. The kernel's approach to caching is optimized for the general case, but each program is specific. It's plausible to me that a caching service like Varnish is a very close match. But that's an empirical question, not a theoretical one.