One of the clients was performing a study and needed help identifying the different factors that would influence the decision made by small medium or large enterprises in choosing between the different options available for kafka.
I did the research and categorized the needs based on whats a must have for small business vs for a large enterprise and which of the available options for kafka provide that.
I have summarized some of the costs for each of the options for equivalent implementations. As always reach out to the Data Experts at Aqib Technologies if you have questions / concerns / comments or need help in any implementation.
Features | Small/Medium business | Large Enterprise | Apache Kafka | Confluent Kafka | AWS Kinesis | AWS MSK |
Serverless | Critical | Not needed | N | Y | Y | Y |
Elastic Scaling | Critical | Nice to have | N | Y | Y | Y |
Infinite Storage / Tiered Storage | Nice to have | Nice to have | N | Y | N | Y |
High Availability | Critical | Critical | Y | Y | Y | Y |
No ZooKeeper management | Nice to have | Nice to have | Y | Y | Y | N |
No-touch patching and upgrades | Nice to have | Critical | N | Y | Y | Y |
Connect to data sources / data sinks | Critical | Critical | Y | Y | Y | Y |
Kafka Streams | Critical | Critical | Y | Y | Y | Y |
Java Clients | Critical | Critical | Y | Y | Y | Y |
Non Java Clients | Critical | Critical | Y | Y | Y | Y |
REST Proxy | Critical | Critical | Y | Y | Y | Y |
Connectors | Critical | Critical | Y | Y | Y | Y |
MQTT Proxy | Critical | Critical | Y | Y | Y | Y |
Schema Registry | Nice to have | Nice to have | Y | Y | Y | Y |
ksqlDB | Not needed | Not needed | N | Y | Y | N |
Cloud UI / Control Center | Critical | Nice to have | N | Y | Y | Y |
Metrics API | Critical | Critical | Y | Y | Y | Y |
Datadog / Prometheus Integration | Nice to have | Nice to have | N | Y | Y | Y |
Health+ | Critical | Critical | N | Y | Y | Y |
Admin REST APIs | Nice to have | Nice to have | N | Y | Y | Y |
Confluent for Kubernetes | Nice to have | Nice to have | N | Y | Y | Y |
Ansible Playbooks | Nice to have | Nice to have | N | Y | Y | Y |
Self-Balancing Clusters | Nice to have | Critical | N | Y | N | Y |
Access Control Lists | Nice to have | Nice to have | N | Y | Y | Y |
Role-Based Access Control | Nice to have | Critical | N | Y | Y | Y |
Audit Logs | Nice to have | Critical | N | Y | Y | Y |
Secret Protection / Encryption | Nice to have | Critical | N | Y | Y | Y |
Bring Your Own Key | Nice to have | Critical | N | Y | Y | Y |
Private Networking | Nice to have | Critical | N | Y | Y | Y |
HIPAA/GDPR/CCPA readiness | Not needed | Critical | N | Y | Y | Y |
Schema Validation | Nice to have | Critical | N | Y | Y | Y |
Stream Catalog | Nice to have | Nice to have | N | Y | Y | Y |
Stream Lineage | Nice to have | Nice to have | N | Y | Y | Y |
99.99% Uptime SLA | Nice to have | Critical | N | Y | Y | Y |
Cluster Linking | Nice to have | Nice to have | N | Y | Y | Y |
Multi-AZ / Multi-Region Clusters | Nice to have | Critical | N | Y | Y | Y |
Replication | Critical | Critical | Y | Y | Y | Y |
24x7x365 Expert Support | Critical | Nice to have | N | Y | Y | Y |
Professional Services | Critical | Nice to have | N | Y | Y | Y |
Education | Critical | Critical | Y | Y | Y | Y |
Monthly Cost in Dollars for 1 MB/Sec | 60 | 750 | 900 | 300 | ||
Performance (MB ./ sec ) | 600 | 600 | 200 | 600 | ||
Ease of Deployment (Scale 1 – 5) | Critical | Critical | 5 | 1 | 2 | 2 |
Ease of Maintenance (Scale 1 – 5) | Critical | Critical | 2 | 1 | 1 | 1 |