I like where this is going as YARN is limiting the Big ecosystem to Java workloads and constraints… but there is so much more to do to get this Enterprise : a good first step would be Kerberos for Authentication to other Hadoop components.

Also, I question the value of Kubernetes on Cloud as these workloads are already generally scaled as required using AWS, GCS or Azure VMs with the Hadoop technology of choice. To me, Kubernetes works better to improve resource management for bare metal private clouds… which doesn’t receive the buzz of public cloud is still approximately 80% of long lived big data clusters.

Source link

No tags for this post.


Please enter your comment!
Please enter your name here