Terence Kelly wrote:
> Recent research suggests
> that policies such as GreedyDual-Size and variants of LFU
> offer substantial advantages over LRU-based algorithms.
I may not be up-to-date, but all improvements of LRU I have seen so far
fail when applied to real cache sizes. Clearly, having caching policy as a
module would benefit a few researchers, but I still have to see a solid
proof that a stupid LRU-TH can be outperformed by a "traditional" smart
algorithm. If you have a pointer to a corresponding article (that uses
caches size more than 300% of high volume daily traffic), please send it to
me.
Alex.
Received on Tue Jul 29 2003 - 13:15:58 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:09 MST