Expanding Apache Kafka Multi-Tenancy for Cloud-Native Systems ft. Anna Povzner and Anastasia Vela |

‌

Episode description

‌

Episode description

In an effort to make Apache Kafka® cloud native, Anna Povzener (Principal Engineer, Confluent) and Anastasia Vela (Software Engineer I, Confluent) have been working to expand multi-tenancy to cloud-native systems with automated capacity planning and scaling in Confluent Cloud. They explain how cloud-native data systems are different from legacy databases and share the technical requirements needed to create multi-tenancy for managed Kafka as a service.

As a distributed system, Kafka is designed to support multi-tenant systems by:

Isolating data with authentication, authorization, and encryption

Isolating user namespaces

Isolating performance with quotas

Traditionally, Kafka’s multi-tenant capabilities are used in on-premises data centers to make data available and accessible across the company—a single company would run a multi-tenant Kafka cluster with all its workloads to stream data across organizations.

Some processes behind setting up multi-tenant Kafka clusters are manual with the requirement to over-provision resources and capacity in order to protect the cluster from unplanned demand increases. When Kafka is on cloud instances, you have the ability to scale cloud resources on the fly for any unplanned workloads to meet expectations instantaneously.

To shift multi-tenancy to the cloud, Anna and Anastasia identify the following as essential for the architectural design:

Abstractions: requires minimal operational complexity of a cloud service

Pay-per-use model: requires the system to use only the minimum required resources until additional is necessary

Uptime and performance SLA/SLO: requires support for unknown and unpredictable workloads with minimal operational workload while protecting the cluster from distributed denial-of-service (DDoS) attacks

Cost-efficiency: requires a lower cost of ownership

You can also read more about the shift from on-premises to cloud-native, multi-tenant services in Anna and Anastasia’s publication on the Confluent blog.

EPISODE LINKS