Apache Pulsar Cluster
Apache Pulsar is a distributed publish-subscribe messaging and streaming platform, originally built at Yahoo, that separates stateless message brokers from durable storage handled by Apache BookKeeper so each layer can scale independently.
Baseline topologies
| Topology | Minimum nodes | Roles |
|---|---|---|
| minimal-cluster | 6 | pulsar-broker, pulsar-bookie, pulsar-zookeeper |
| scaled-cluster | 9 | pulsar-broker, pulsar-bookie, pulsar-zookeeper |
What drives the cost
- Peak throughput (MB/s)
- Peak aggregate produce throughput in megabytes per second. Drives broker CPU and bookkeeper write throughput requirements.
- Total topic/partition count
- Total number of topics and partitions. Drives broker memory for managed ledger caches and bookkeeper journal parallelism.
- Total retention storage (GB)
- Total data retention across all topics. Drives bookkeeper ledger storage capacity; tiered storage offload can reduce hot-tier requirements.
Where it runs
bare_metal, dedicated, cloud_vm
Size Apache Pulsar Cluster and get a cost estimate