Kubernetes or ZooKeeper for HA?

[mikep]

Hello. I am trying to understand High Availability in DataBricks. I understand that DB uses Kubernetes for the cluster manager and to manage Docker Containers. And while DB runs on top of AWS or Azure or GCP, is HA automatically provisioned when I start a cluster because of Kubernetes or Does it use ZooKeeper? In another scenario, if I configure the cluster mode to Standard, I only have one Driver (Master Node) which can be a single point of failure. Will Kubernetes take care of it if it fails or does it make sense to start up ZooKeeper to maintain a Quorum? Or does Kubernetes by default use ZooKeeper to watch ALL nodes? I hope this makes sense.

[Kaniz]

Hi @mikep! My name is Kaniz, and I'm the technical moderator here. Great to meet you, and thanks for your question! Let's see if your peers in the community have an answer to your question first. Or else I will get back to you soon. Thanks.

[mikep]

Thanks so much Kaniz. Ultimately, I'm looking for any architecture details on how DB configures Kubernetes to manage nodes. Especially the Driver (Master Node). It's clear that Apache Spark can be configured to use ZooKeeper but I don't find anything regarding DataBricks. Thanks for your time.

[Kaniz]

Hi @Mike Poullas​ , Perhaps this link might help you.

Please watch this video too.

Let me know if this helps.

[Anonymous · Accepted solution]

Hey @Mike Poullas​ , regardless of any cloud providers, Cluster Manager ensures HA for your driver node.

Under the hood, almost all Databricks services are running on some K8s cluster.

ZooKeeper is not used.

[Kaniz]

Hi @Mike Poullas​ , Did @Ivan Tang​ 's reply answer your query?

[Anonymous]

https://buyusasmm.com/product/buy-google-5-star-reviews/