Clustering

Stateless Active/Active Clustering

Kafkorama can be deployed as a stateless cluster composed of multiple independent Kafkorama Gateway nodes, with Kafka serving as the communication backbone between nodes. Because no user state is shared across the cluster, Kafkorama scales horizontally and linearly, both in terms of subscribers and publishers.

The stateless design also simplifies cluster management in cloud environments by enabling dynamic scaling through cloud-native tools such as Kubernetes.

Kafkorama clustering provides:

• Fault tolerance with no single point of failure
• Horizontal scalability through native load distribution

As a result, a Kafkorama cluster inherently supports load balancing and fault tolerance, eliminating the need for a dedicated load balancer or application delivery controller (ADC). However, deploying Kafkorama behind a load balancer or ADC may still be beneficial in certain enterprise environments, depending on the customer ecosystem.

Cluster QoS

Kafkorama supports two clustering modes, corresponding to different qualities of service (QoS):

Quality of Service	Clustering Type
Standard	Standard Message Delivery
Guaranteed	Guaranteed Message Delivery

For details about these QoS levels, see the QoS section.