how to optimize the time of retrieving data out of the trino

[zeddit]

I want to use trino to get data from multiple backend, and the using the data to do some data science in jupyter notebook.

however, I found the speed to acquire data is quite slow. e.g. it could reach less than 15MB/s while I have a 10Gb/s network.

I use the memory connector to find out the bottleneck and the trino worker could talk with my backend database at a speed about more than 100MB/s.
because I found the bottleneck is at trino itself, rather than the database, so it cannot be optimized with the help of alluxio and caches like it.

so is there any method to make it faster, and what's the underlying bottleneck in trino that limit the speed. great thanks.