You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@lucene.apache.org by Apostolis Xekoukoulotakis <xe...@gmail.com> on 2012/06/12 12:42:34 UTC
A simple question on disk IO and posting lists disk format.
Lets suppose that we make a query with multiple terms. Lucene creates a
topScoreDocsCollector with an Inorder traversal of posting lists.
Lets suppose we are in a specific segment, since we use a Priorityqueue in
the topScoreDocsCollector, I assume that all posting lists are traversed
concurrently.
Does lucene use a buffer to reduce disk seeks? what is its size? or does
lucene load all the posting lists into memory?
The second seems more plausible.
(I ask because as I said in a previous message, I am creating a
TopDocsCollector with multiple PriorityQueues and external "posting
lists"(in fact they are ordered scores) from a database(levelDB))
--
Sincerely yours,
Apostolis Xekoukoulotakis