Most optimized way to keep the libraries loaded ?

[Mazenode]

Hi,

I want to create a realtime API using Dataiku, the thing is that the API will be called a lot per day, but the libraries required for the treatment takes very long to load (about 20 seconds). Is there a way to keep the librairies permanently loaded in the DSS ? So that I only need to run my ML/ DL jobs without having to reload them everytime.

Thanks by advance 

[Clément_Stenac]

Hi,

If you create a Python endpoint using the DSS API Designer / API node, the libraries will stay loaded between calls:

https://doc.dataiku.com/dss/latest/apinode/index.html 

Best,

[Mazenode]

Thank you for your answer, it simplifies a lot of things for me ! 😄

Best,