Speaker
Description
At Uninett, we have chosen Kubernetes as a platform to provide various
services. We have currently two production clusters serving different
communities. In addition we have used public cloud resources through kubernetes
for some use cases as well.
To make use of platform easier by end users, we have made an application
store where users have commonly used tools such as Jupyter Notebook,
Jupyter Hub, Rstudio, Spark to process their stored data. Users can
customize these tools with the packages they need and run it with the
same ease. We also provide Minio as sync &share solution due to its
simplicity to manage and operate it.
Deep learning is another area where our users have shown interests and
to address those needs, we provide an application to enables use of
most common frameworks e.g. Tensorflow, Pytorch, Keras etc on a
GPU enabled infrastructure. User can request multiple GPUs to scale
their workflow as needed.
This presentation will go through our platform and application store. I
will also share experience from our users using these application for
various use cases.
