US 12,248,473 B1
Query performance prediction using multiple experts
Zhengchun Liu, Sunnyvale, CA (US); Gaurav Saxena, Cupertino, CA (US); Balakrishnan Narayanaswamy, San Jose, CA (US); Kaihui Zheng, Mountain View, CA (US); Mohammad Rezaur Rahman, Fremont, CA (US); and Tim Kraska, Arlington, MA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Dec. 14, 2023, as Appl. No. 18/540,496.
Int. Cl. G06F 16/2453 (2019.01)
CPC G06F 16/24545 (2019.01) 20 Claims
OG exemplary drawing
 
5. A method to predict one or more performance characteristics of a query, comprising:
analyzing a query plan of the query to identify a feature vector of the query;
responsive to determining that the feature vector matches respective feature vectors of a number of previous queries in a history of feature vectors, the number exceeding a threshold value:
predicting the one or more performance characteristics of the query according to respective performance characteristics of one or more previous queries identified in the history having the feature vector; and
responsive to determining that the feature vector does not match respective feature vectors of a number of previous queries or matches respective feature vectors of a number of previous queries, the number not exceeding a threshold value:
predicting the one or more performance characteristics of the query according to a machine learning model trained according to respective performance characteristics and feature vectors of a plurality of queries.