Complimentary technologies are very important as it's rarely possible for a single technology to be optimized for everything. There are always trade-offs - jack of all trades master of none also applies in technology solutions: https://www.graphgrid.com/jack-of-all....
In this increasingly polyglot world of data storage options one natural pairing that has proven quite effective is Neo4j and ElasticSearch. Neo4j is designed for optimal graph traversal (i.e querying highly connected data) and ElasticSearch is a powerful search server with great language analyzers and aggregation strategies out of the box. While you could try to rebuild the search capabilities in Neo4j because it's built on Lucene, you'd be relaying much of the foundation you already have in ElasticSearch with a subpar result and similarly while you could try to represent connected graph structures in ElasticSearch you'd be expending much effort with short comings in reliability due to the underlying storage mechanisms compared to using Neo4j for that connected data aspect. In this talk we'll look at paring them together and reaping the benefits of both doing what their best at.
In this increasingly polyglot world of data storage options one natural pairing that has proven quite effective is Neo4j and ElasticSearch. Neo4j is designed for optimal graph traversal (i.e querying highly connected data) and ElasticSearch is a powerful search server with great language analyzers and aggregation strategies out of the box. While you could try to rebuild the search capabilities in Neo4j because it's built on Lucene, you'd be relaying much of the foundation you already have in ElasticSearch with a subpar result and similarly while you could try to represent connected graph structures in ElasticSearch you'd be expending much effort with short comings in reliability due to the underlying storage mechanisms compared to using Neo4j for that connected data aspect. In this talk we'll look at paring them together and reaping the benefits of both doing what their best at.
No comments:
Post a Comment