Using openstack for training machine-learning models in the cloud?

asked 2018-02-08 12:08:37 -0600

Median gravatar image

Hello everyone. I am very new to anything related to cloud in general, so I apologize if this question is newbish or does not comply with the forum rules.

I am working on a project that requires heavy machine learning models to be trained. There are some features that we would like to implement that we think cloud computing/storage could provide us, and I would like to know if OpenStack could be the way to go.

1 - Training these machine learning models could take days or weeks on a standard machine. Is it possible to use openstack to take advantadge of remote CPU/GPU resources to make this process faster?

2 - At times, it may be necessary to provide software updates for the users, and these updates may require downloading some models (.txt / .csv / .xml files) into user's products. Can this be achieved using openstack?

3 - Can openstack be used to create and mantain a storage cloud for an image database (.png / .jpg / .bmp), and have this database be accessed by our software for retrieval and processing?

If OpenStack is the way to go for these kind of features, can you provide some resources for me to start get acquainted to how these things are done in practice? I've been trying to read some resources about this topic, but everything just sounds so abstact that I'm feeling a bit lost.

Thanks in advance.

edit retag flag offensive close merge delete