I'm using Elasticsearch, and for paginating the search results, should I use the simple from/size, or should I go with search after. I've read that if the number of hits is less than 10k, then from/size is acceptable. Even though my total search space is much more than 10k, but the number of hits/matches I'm expecting is around 5k at max.
For this, would there be a lot of performance difference between from/size and search after? I'm leaning towards from/size since search after is relatively trickier to implement, specially when there's a requirement for going back to a previous page.
Hi,
Thank you for your response, and the helpful links!
But is there some threshold/recommendation as to what would be the maximum size of the result set on which from/size and search after would perform roughly the same?
I understand that it will also depend on other factors, like query logic, or size of the analysed data we're querying on(The size of 1 document for us is about 5-10kb).
We're building a e-commerce type search from scratch, and I don't expect the search results of a query to be more than 500, or 1k.
It's difficult to to do a POC on both approaches, since we don't have data in that scale yet, and I'm leaning towards from/size. Hence, was looking for some empirical data/advice from people with more experience on this.
It seems that one of the requirements is to have access to the previous page, in which case I would also use from/size respecting the restrictions of this implementation.
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.