On Thu, Apr 17, 2014 at 6:50 PM, Greg Stark <stark@mit.edu> wrote:
> On Thu, Apr 17, 2014 at 4:48 PM, Peter Geoghegan <pg@heroku.com> wrote:
>> Although I will add that not caching highly useful inner pages for the
>> medium term, because that index isn't being used at all for 5 minutes
>> probably is very bad. Using the 4,828 buffers that it would take to
>> store all the inner pages (as in my large primary index example) to go
>> store something else is probably penny wise and pound foolish.
>
> But there could easily be 20 unused indexes for every 1 index that is
> being used.
Sure, but then there might not be. Obviously there is a trade-off to
be made between recency and frequency. One interesting observation in
the LRU-K paper is that for their test case, a pure LFU actually works
very well, despite, as the authors acknowledge, being a terrible
algorithm in the real world. That's because their test case is so
simple, and concerns only one table/index, with a uniform
distribution.
--
Peter Geoghegan