Azure Search Best Practices RRS feed

  • Question

  • I am looking for Azure Search best practices to answer the following question:

    What are the pros/cons of using Azure Search independently(without backing it with an actual DB, like DocumentDB)?  From what I can tell, an actual DB is NOT required and I'm leaning towards that direction.  However, since I'm new to Azure Search, I want to see if that is a best practice or an anti-pattern.



    Wednesday, February 3, 2016 3:07 PM

All replies

  • Whether or not it's a good idea to use Azure Search as a primary data store depends on your needs and access patterns. Azure Search is great for full text search - it is not as good as, say, Azure SQL or DocumentDB for other data access scenarios such as complex aggregations, joins, BLOB storage, etc. If your data exists to primarily drive search then just Azure Search is fine; if you need other access patterns I mentioned, customers often keep primary data in Azure SQL or DocumentDb, and index a subset of that data necessary for search into Azure Search. Our indexers (https://azure.microsoft.com/en-us/documentation/articles/search-indexer-overview/) make it simple to move the data from those places to search.


    Thanks! Eugene Shvets Azure Search

    Wednesday, February 3, 2016 8:12 PM
  • Hi Jim

    I'd recommend against this approach for a simple reason, right now you cannot edit Index fields, you can edit an Index and add fields, but what if you need to change a field type? Or change a field and make it Searchable? You'd lose all your data unless you exported it.

    The only way of doing it is if your data is stored elsewhere and you can recreate / reindex your items with the new data.

    Implementing Indexers over your data is really simple, I'd recommend that approach and using DocumentDB or Azure SQL, depending on your data (or Blobs if it applies).

    Thursday, February 11, 2016 2:12 AM