Open lucasjinreal opened 5 years ago
@jrruijli any hints? download via aws cli is indeed very slow. Thank you very much.
I was able to get a reasonable speed (~1.1 MiB/s) by replacing sync with cp:
aws s3 --no-sign-request cp --recursive s3://open-images-dataset/train train
aws s3 --no-sign-request cp --recursive s3://open-images-dataset/validation validation
aws s3 --no-sign-request cp --recursive s3://open-images-dataset/test test
Explanations here.
You will need to have a steady internet connection, because if it drops and you restart the cp, it will restart from the beginning.
Also, keep in mind that aws s3 can quickly reach your hosts limits if you use the default throttling settings (it happened to me on several occasions). I suggest that you take a look at this to configure it.
Best luck !
@ltrottier thank you very much, that was helpful!
Completed 256.0 KiB/45.9 GiB (3.2 KiB/s) with 1 file(s) remaining
It's need about 100 years to finish in current speed