lucene-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <>
Subject Re: indexing on nas storage
Date Fri, 20 Mar 2009 01:38:07 GMT
There are no issues running lucene on any drive that provides fast and
reliable random access reads.

Some SAN drives will work better than cheap local disks and those work
pretty well.

It is even possible to run Lucene with an index in a decidedly unfriendly
file system (from the standpoint of random access reads) like HDFS.

How well it works depends a lot on your particular work load.  The long tail
applies here; most retrieval applications are pretty small and only a few
are really, really huge.  For small applications up to a million or a few
million documents and queries arriving every few seconds, and low update
rates, you should be fine almost no matter what you are using.  For hundreds
of queries per second against hundreds of millions of documents with lots of
updates, you have a completely different kettle of fish that will require
completely different techniques.  For really large systems, you have to
implement scalable clustered systems and the necessary considerations are
much broader than just disk I/O rates.

On Thu, Mar 19, 2009 at 5:58 PM, 이지홍 <> wrote:

> I wonder if there are any known issues having a lucene index on a NAS
> or SAN drive?  Some
> basic tests show that it works fine.  But are there performance issues
> with indexing on NAS
> for instance?
Ted Dunning, CTO

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message