Words vs Guids, what is faster?

[Baygon (Vincent T)]

I'm in front of a design choice where I would need a bit of help to choose the most efficient way.

I'm currently indexing data tagged with a string of comma separated Guids, for example:
\"e659959d-392f-44c5-83a5-fb959cdbaccc\",\"ab2975e3-b9ca-4b1a-a93e-fb61a5d5c3a4\",\"c48e0bf4-7a1b-4ffd-893c-12e46e664f7f\",\"0074af4d-eb56-4e57-89b6-07c39c63c9c4\"

This is the exact look of the string which is in one field only (ingestion in logstash sql of a serialized list stored in the mysql db).

I can modify the logstash query to translate the Guids to english words, for example:
english, cute, irl

Which one of the above solution would yield the best search performance when executing the following:
GET myindex/_search { "query": { "bool":{ "must":[ {"bool":{"should": [{"match":{"tag_ids":{"query":"ab2975e3-b9ca-4b1a-a93e-fb61a5d5c3a4","operator":"AND"}}}]}} ] } } }

or of course

GET myindex/_search { "query": { "bool":{ "must":[ {"bool":{"should": [{"match":{"tag_ids":{"query":"english","operator":"AND"}}}]}} ] } } }

Complexity of ingestion is minor while search performance is critical. Thanks for your inputs

[warkolm (Mark Walkom)]

The best approach here is to test each and see what gets you the results you want.

[Baygon (Vincent T)]

not really, it would take about 2 weeks to ingest the data, ie 4 weeks to reach a conclusion, so if someone knows the logic behind the fuzzy search and how words/guids are indexed and how it affects performance that would be valuable, not only to me, but to anyone trying to design a performant ES index