Item: Elasticsearch
Rating: 9
Author: Verified User

Overall Satisfaction with Elasticsearch

Use Cases and Deployment Scope

We get a lot of scientific journals in pdf format. Windows only allows title search. Some scientists use Mendeley but there is a licensing cost involved. We implemented Elasticsearch to help the scientists to search by author or look for keywords in the title or in the content. And we have provided options to look for an exact match as well as partial match.

Pros and Cons

Pros

The snippet that we get back before and after the search words is very helpful for the scientists to get the right content.

Cons

At my previous job with a simple 3 node cluster, Elasticsearch did not do a good job in electing a new master, when the master node went down. Many times, I had to stop and restart all the nodes to make it function again. I have heard implementation models where customers had dedicated multiple nodes just for master.

Return on Investment

Apart from the operational issue that I mentioned previously, the business users have been extremely happy with the results as well as the short time taken to perform the search.

Alternatives Considered

Apache Solr

Ability to support JSON queries, Percolator, ease to set up and custom routing were some of the reasons why we decided to use Elasticsearch instead of Solr.