Speed and size difference with "mult-fields, multi-analyzers" Indexing


I just reviewed the documentation for the fields option:

I plan on indexing my data with a snowball analyzer and also retaining the regular version of a field with a text type.

I'm wondering if using this approach would double the size of the index? With the original copy and the "snowball analyzed" copy?

Also, does querying the "fields" or "embedded" version have any significant query time speed slow downs? It reminds me of the nested query, so I'm nervous it might slow down the overall search/query speed.


Any ideas? Hoping someone else has implemented this approach and can share their experience

That's probably the right way to solve your use case.


I am a heavy user of multi-fields, yes the size of the document increase and it depends on your document, if you have only 1 field and add analyzer to this field there's a chance to double, depends on the result of the analyzer I think...

For speed concern I didn't notice difference. Hope that my post help you to feel less nervous. :grin:

Depends on how much documents you have, so better to make some benchmark with your data before.

Thank you! I'll keep all of it in mind

Just wanted to update. I made 3 of my fields into "multi-fields" with a snowball analyzer and just a regular "text" type copy.

Obviously, your own results are going to be different, but I thought I'd share my stats regardless.

This is a small subset of my data, but here's my dataset stats the old way (all 3 fields only had a snow-ball analyzer ... no text copy):

After re-ingesting the data as multifield:
It was about double after the multifield mapping, which is what I was expecting.

However, after about 15 minutes, the size of the index went down:
Chart confirms it ...

nice to see elasticsearch doing its magic! Appears to have taken less than double in my case, which is awesome to see.

1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.