lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Erick Erickson <erickerick...@gmail.com>
Subject Re: Need Debug Direction on Performance Problem
Date Sun, 18 Jan 2015 00:08:27 GMT
Hmmm, you say
"reading around 10 lakh docs". Are
you returning 1,000,000 documents? That is,
is have you set &rows=1000000? Returning
that many rows will never performa all that well. What
kinds of performance are you getting if you only
are reading a few rows?

Or have I misunderstood completely?

Best,
Erick

On Fri, Jan 16, 2015 at 8:22 PM, Naresh Yadav <nyadav.ait@gmail.com> wrote:
> Hi all,
>
> We have single solr index with 3 fixed fields(on of field is tokenized with
> space) and rest dynamic fields(string fields in range of 10-20).
>
> Current size of index is 2 GB with around 12 lakh docs and solr
> nodes are of 4 core, 16 gb ram linux machines.
>
> Writes performance is good then we tested one read query(In select query we
> are applying filter criteria on tokenized field & reading only score field,
> no grouping/faceting) in two setups :
>
> *Setup1 : *Single Node Cloud with shards=1, replication=1
> In this setup whole 12 lakh docs are on same machine. Our filter query
> reading
> around 10 lakh docs with only score field is taking *1 minutes*.
>
> *Setup2 : *Two Node Cloud with shards=2, replication=1
> In this setup whole 6 lakh docs on node1 and 6 lakh on node2. Our same
> filter query reading around 10 lakh docs with only score field is taking *114
> minutes.*
>
> Please guide us what can be possible reasons of degradation of
> performance after sharding of index. How can we check where solr server
> is taking time to return results.
>
> Thanks
> Naresh

Mime
View raw message