page cache studies
Hello,
I'd like to ask, are there any numerical studies of current Linux page cache implementation? I don't mean those nice design documents but rather theoretical estimates based on it, e.g working set size, active and inactive list sizes and cache hit probability one should expect for a given workload as a stochastic process. I suppose, there is no much use of it, but it would be rather curious reading anyway.
Thanks in advance.
Regards,
Vladimir