AutoMQ adopts a storage-compute separation approach, evolving Apache Kafka® from a Shared Nothing architecture to a Shared Storage architecture. This shift enables full utilization of cloud-native technology and cost advantages. Technically, the shared storage architecture offers inherent benefits in addressing operational challenges like partition reassignment and automatic scaling. AutoMQ leverages these benefits through product capabilities, as shown in the table below.
AdvantagesAutoMQApache Kafka®
Native support for Kafka protocol✅ Natively supported, reuses protocol layer code✅ Natively supported
Second-level partition reassignment✅ Shared storage, no need to copy data during reassignment❌ Reassigns local data, typically takes hours per partition
Second-level smooth scaling✅ Achieves second-level scaling by bulk reassigning partitions❌ Due to data copying during partition reassignment, cannot support fast scaling, typically takes tens of hours
Continuous self-balancing✅ Built-in Auto Balancing component, dynamically adjusts traffic to eliminate hotspots❌ Prone to local hotspots, affecting stability
Stateless Broker✅ Data offloaded to shared storage, Brokers are stateless and can use Spot instances❌ Local disk state, requires extensive data transfer for decommissioning
Cold-hot isolation, 5x cold read efficiency✅ Out-of-the-box cold-hot isolation capability, supports multi-tenant business scenarios❌ Cold and hot data affect each other, exacerbating system degradation, cannot support multi-tenant scenarios