Is Lucene updateDocument faster than deleting and then adding document?

Question

I have a large index (about 100 GB) and I want to update documents in the index frequently. I'm in doubt between 2 methods:

1) Updating the document

2) Deleting the document and adding the updated version

Which one would be faster? Is there any other pros and cons!?

Answer 1

Regarding the Lucene API documentation, there should be no difference between updating a document or removing the old and adding the new one. Internally updating causes a remove and add operation:

In either case, documents are added with addDocument and removed with deleteDocuments(Term) or deleteDocuments(Query). A document can be updated with updateDocument (which just deletes and then adds the entire document). When finished adding, deleting and updating documents, close should be called. ( http://lucene.apache.org/core/4_6_0/core/org/apache/lucene/index/IndexWriter.html )

Answer 2

If you can batch your deletes and adds, the best practice is to first make all deletes and then do all adds. My tests on large indices proved that to me.