cancel
Showing results for 
Search instead for 
Did you mean: 

How cost effective is creating full text index for Fuzzy search?

Former Member
0 Kudos

Hello Team,

I have a requirement to have a fuzzy search on 1 each column of 3 huge tables (2 billion, 1 billion and 0.5 billion each). SAP recommends creating full text index for those columns with NVARCHAR datatype as we are not able to modify existing table structure to SHORTTEXT. Have anyone tried creating full text index and did analysis on how cost effective it is to create? Memory consumed by table shoots up by 2X if i create Full text index on only 1 column. Is it the expected behaviour? How much is the gain in search performance is observed here?

Thanks,

Akshay

Accepted Solutions (0)

Answers (1)

Answers (1)

former_member182302
Active Contributor
0 Kudos

Hi Akshay,

Well if SAP is recommending, i will surely go with them

Regarding your question, is the requirement is to tokenize every string with space delimeter or you have some rules/criteria and you want the data to be stored with only tokens in the index?

Regards,

Krishna Tangudu

Former Member
0 Kudos

Hi Krishna,

I have both requirement for token search as well as whole word search. Both are different tables. What is the optimal way of creating index?

Thanks,

Akshay

former_member182302
Active Contributor
0 Kudos

Hi Akshay,

Whole word search you can use dictionary and for the token pattern search you can use CGUL rules and filter the categories you want using the config file . This should help the maintain the volume and you can partition and distribute over the nodes in your system, if you intend to have data balancing.

Regards,

Krishna Tangudu