MDM Performance

[Former Member]

Hi All,

We are experiencing problems with matching records in MDM. The database contains ~500k records and when trying to match even 1 record vs all using some fairly simple matching strategy it takes 40 minutes.

Does anyone have any similar experiences and any advice on how to improve the performance of the matchings? The strategy and rules seems fairly optimized. Is MDM qualified for matching this large number of records?

Kind Regards,

Christian

[Former Member · Accepted solution]

Hi christian

I can share some of the guidlines for Matching strategy in MDM for optimum performance.

1> Use equals rather than Token equals wherever possible.

2> For Undefined scenario also put a score. When NULL is matched with some value for a field.

3> In large no of records try to run Selected Vs result for the matching Strategy. Try to reduce the total no of records on which you want to run the matching strategy by a search and save this as named search.

this will surely take care of some of issues.

Please Award if helpful.

Best regards

Ravi Kumar

[Former Member]

Hi Christian,

The records in use for finding match duplictaes are large though but not as much to cause such a poor performance .A few steps can be taken to better the performance

- Since you are comapring records within the MDM repository itself you have a fair idea of the records and the major duplictaes.Try minimizing the matcing time by using transformation wherever possible.by this you can eliminate a lotof minor mistakes like space and special character like difference between rec

- Also when you use transformation you can use the token equls instead of only equals this will futher reduce the processing time.

- Identify the primary field which will be participating in the duplicates finding and use minimum possible rules but in an effective strategy

Hope It Helped,

Thanks & Regards

Simona Pinto

[Former Member]

Christian,

In addition Influencing parameters for matching time are: Number of matching rules, source records, target records.

Also work on ur Transformation since its going to be string comparison, minimize or use most appropriate Transformations.

Anway bottom line is, these will not improve the performance drastically but some extend.

thanks

Alexander