Detection of Performance-Diminishing Match Rules

You need to detect performance diminishing match rules to ensure that you can take corrective action.


You must take care in designing the tokenization scheme for a match group. A Token Phrase is the concatenated combination of tokens generated for an entity by a rule. For example, (<alex:yg-12:ca:91301>). The generation of token phrases for an entity can become excessive due to the combination of attributes within the rule that are mapped to token classes and the number of values contained within those attributes for a set of records. Certain limits are imposed to protect the performance of your tenant and its surrounding services.

Operation of the performance-diminishing match rule detection service

The performance-diminishing match rule detection service executes as token phrases are generated by the match engine for a given entity. There is an internal limit of 300 token phrases per entity. When 100 entities exceed this limit, then the customer administrator assigned to the tenant will receive an email notification advising that the performance of the tenant is likely diminished. The offending match group(s) are listed in the email.

Remediating an offending match rule

If you modify the configuration of an offending match rule (other than merely changing its scope), then the rule is considered safe again until such time in the future that tokenization potentially detects it as being a performance-diminishing rule again and flags it accordingly.

Note: The Match Rule Analyzer inspects the match rules and provides suggestions to improve the performance of these rules. A match analyzer can detect and inspect the malicious behavior of match rules. For example, overcollisioned match tokens lead to large amounts of matches, which cause many comparisons that significantly load the platform. Popular token phrases and corresponding entities are detected and notifications are triggered or other actions, such as match rule disablement are invoked. The proactive notifications include information about entity types and match rules.