We use apache-solr with over 10 million documents with a merge factor 2. It is hosted on amazon ec2 (amazon linux) with ebs volumes attached.Index resides on ebs.During merge, the io limit is reached and causing issues.
Is there any way to limit iops at process level or at particular disk level?
or can we slow down solr merging?
Thanks
We use apache-solr with over 10 million documents with a merge factor 2. It
Share
You can mount several EBS in a RAID volume to get a higher throughput. This post from Heroku provides interesting facts of EBS io performance.
It’s not a direct answer to your question but it will push the limit (if you are not already using RAID).