Newbie with Kubernetes and infrastructure stuff and hence asking this question. Is it possible to launch Pinot components in multiple AWS EC2 instances like we do in a multi cluster druid deployment? Are there any drawbacks to this approach except for the obvious difficulty in managing multiple servers?

Hi Rohit, Yes it is possible to launch Pinot components in multiple EC2 instances. Just follow this thread https://apache-pinot.slack.com/archives/CDRCA57FC/p1664035178017179

@Ganesh few more doubts 1. the schema, ingestion spec etc are to be run on the ec2 instance that has the controller running? 2. As a follow up to the above question, if its yes, and we have multiple controller instances, do we run the schema and ingestion spec on each of the controller instances? 3. If one wants to have multiple zookeeper instances, do we provide all the zookeeper instance addresses as -zkAddress parameter (maybe as comma separated values?) The architecture i have in mind is 3 zookeepers 2controllers 3 brokers 3 servers

Typically you’d have a load balancer in front of controllers or Zk and you provide that

So i am assuming one cannot provide multiple zookeeper addresses using the zkAddress parameter and will have to use a loadbalancer ? Also, regarding the specs, do we run the schema, ingestion spec on all the ec2 instances having controller? Just trying to wrap my head around this.