Performance impact of wildcard search

To optimize performance of wildcard search, use the following recommendations.

Account for increased time needed for indexing. In general, if wildcard search is enabled in Endeca Server (even if it is not used by the users), it increases the time and disk space required for indexing. Therefore, consider first the business requirements for your Endeca application to decide whether you need to use wildcard search.
Note: To optimize performance, the Dgraph performs wildcard indexing for words that are shorter than 1024 characters. Words that are longer than 1024 characters are not indexed for wildcard search.
In addition, if the wildcard search is not enabled before you load the records, and you issue a Configuration Web Service request to enable it, after a large number of records exist in the data domain, this causes re-indexing and is associated with increased processing time.
Do not use "low information" queries. For optimal performance, use wildcard search queries with at least 2-3 non-wildcarded characters in them, such as abc* and ab*de, and avoid wildcard searches with one non-wildcarded character, such as a*. Wildcard queries with extremely low information, such as a*, require a significant amount of time to process.
Queries that contain only wildcards, or only wildcards and punctuation or spaces, such as *. (star followed by period), or * * (star space star), are rejected by the Oracle Endeca Server.
Analyze the format of your typical wildcard query cases. This lets you be aware of performance implications associated with one specific wildcard search pattern.

Do you have queries that contain punctuation syntax in between strings of text, such as ab*c.def*?
For strings with punctuation, the Dgraph processes of the data domain generate lists of words that match each of the punctuation-separated wildcard expressions. Only in this case, Dgraph processes of the data domain use the --wildcard-max <count> setting to optimize their performance. Increasing the --wildcard-max <count> improves the completeness of results returned by wildcard search for strings with punctuation, but negatively affects performance. Thus you may want to find the number that provides a reasonable trade-off. You can specify this flag for the data domain profile, with endeca-cmd --put-dd-profile --wildcard-max <value>.