Hi!
We are using Lucene to search a rather large number of docs. We have around
2 million docs now, and it is expected to grow to 50 millions.
We need to sort a lot of documents by various numeric fields, and we can
have 2 sorting fields.
I was told that sorting is slow in Lucene, so we built an index for each
sorting order and the sorting is done by DocId.
We have another index just for searching, where the sort by relevancy.
The results are topped to the first 10,000; we use pagination to go through
them.
The problem is that we are building multiple indexes for each sort order,
and each index is built in a single threaded fashion because we need to
maintain the same order for DocId.
Do you recommend doing something else here ?
I'm using Lucene .net.
--
View this message in context: http://lucene.472066.n3.nabble.com/Sorting-large-indexes-tp4170291.html
Sent from the Lucene - General mailing list archive at Nabble.com.
We are using Lucene to search a rather large number of docs. We have around
2 million docs now, and it is expected to grow to 50 millions.
We need to sort a lot of documents by various numeric fields, and we can
have 2 sorting fields.
I was told that sorting is slow in Lucene, so we built an index for each
sorting order and the sorting is done by DocId.
We have another index just for searching, where the sort by relevancy.
The results are topped to the first 10,000; we use pagination to go through
them.
The problem is that we are building multiple indexes for each sort order,
and each index is built in a single threaded fashion because we need to
maintain the same order for DocId.
Do you recommend doing something else here ?
I'm using Lucene .net.
--
View this message in context: http://lucene.472066.n3.nabble.com/Sorting-large-indexes-tp4170291.html
Sent from the Lucene - General mailing list archive at Nabble.com.