I have around 6 GB of data coming to elastic search per day.
The problem is that it takes around 5-6 seconds to load discover or conduct any search. I was trying to tune - to make it faster.
I have the following observations :
I have enabled the default shard and replica allocation. That is , 5 shards and 1 replica per index. But because I am using one node , the replica is not assigned and hence health of the index is yellow. I can also see that i have 5 unassigned shards. Can this slow down the discover ? The shards are already unassigned , how can it consume resource ?
I don't want to disable the highlight of search.
Also the default discover:sampleSize has been reduced to just 100.
I have done most of the things mentioned already. I have increased the jvm heap to 30Gb from the default of 1 Gb for elastic search (as can be seen in the above scrrenshot). Have got around 15 fields only. If the shards have been unassigned , I dont need to worry that the extra shards are consuming more resources , I guess .
I have a field which I am sorting for each search query . I suspect this is taking up the time. Let me pre-index and sort it through template as mentioned one of the article. I will update here .
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.