Reputation: 4577
In my current (local) deployment of an application, I have Apache Kafka and Zookeeper deployed.
Now I want to put my application in the cloud, and see that AWS MSK is an option. This (currently, at least) uses zookeeper in the background still.
I use Zookeeper, not just for Kafka, but also for leadership election and service discovery (via the Curator set of libraries). Could I use the Zookeeper deployment that comes with MSK to do this still, or would I have to deploy my own Zookeeper ensemble alongside the one that is provided for MSK?
Upvotes: 0
Views: 843
Reputation: 391
@simonalexander2005
It's not recommended to use ZK nodes provided with MSK for any other usage. ZK nodes use tiny instances that are sufficient for MSK, but can be affected by additional workloads, number of connections, frequency of operations, etc.
In other words, you can affect MSK stability.
Also, please note that MSK Serverless doesn't expose ZK nodes at all.
All that being said, you can access ZK nodes using regular ZK client. Connection string can be obtained using MSK APIs.
You will also need to ensure security groups have inbound rules to allow access from ZK client instance.
Upvotes: 2