Expected Behavior
When using GCP Redis as the online store and with 5 feature views, the response time of the get_online_features interface is less than 10ms, specifically 7.27ms 。Official website performance-characteristics
Current Behavior
Using GCP Redis as the online store and with 5 feature views, the response time of the get_online_features interface is over 45ms ,the event_digital_store_push interface is over 20ms ,the event_payment_store_push interface is over 70ms.
Steps to reproduce
In the Python environment, utilize FastAPI to invoke the Feast interfaces for event/payment, /features, /push/payment
online_store:
type: redis
redis_type: redis
connection_string: "redis1:26379,ssl=true,password=my_password"
key_ttl_seconds:604800
Specifications
- Version:0.63.0
- Platform:linux
- Subsystem: registry gcp postgresql
Possible Solution
How to set or adjust the parameters to achieve the 5 Feature views and 7.27 ms response time。
Expected Behavior
When using GCP Redis as the online store and with 5 feature views, the response time of the get_online_features interface is less than 10ms, specifically 7.27ms 。Official website performance-characteristics
Current Behavior
Using GCP Redis as the online store and with 5 feature views, the response time of the get_online_features interface is over 45ms ,the event_digital_store_push interface is over 20ms ,the event_payment_store_push interface is over 70ms.
Steps to reproduce
In the Python environment, utilize FastAPI to invoke the Feast interfaces for event/payment, /features, /push/payment
online_store:
type: redis
redis_type: redis
connection_string: "redis1:26379,ssl=true,password=my_password"
key_ttl_seconds:604800
Specifications
Possible Solution
How to set or adjust the parameters to achieve the 5 Feature views and 7.27 ms response time。