|Lightbend Launches Distributed Cluster|
|Written by Kay Ewbank|
|Tuesday, 16 May 2023|
Lightbend, the company behind Akka, has launched Akka Distributed Cluster. Lightbend produces cloud-native microservices frameworks while Akka Cluster allows for building distributed applications.
Akka Cluster provides a fault-tolerant decentralized peer-to-peer based Cluster Membership Service with no single point of failure or single point of bottleneck. It does this through the use of gossip protocols and an automatic failure detector.
Lightbend says Akka Distributed Cluster provides new capabilities for accelerating data delivery to users, and maintaining application availability even in the event of a cloud provider outage.
One of the difficulties of edge computing is the cost and increased latency of transferring data to the cloud for processing.
Akka Distributed Cluster uses a brokerless publish/subscribe model over gRPC (general-purpose Remote Procedure Call) with low-latency, high-performance and guaranteed delivery so there's no need for message broker management in memory-constrained environments.
The use of gRPC for point-to-point communication with an event journal on the producer side and Akka Projections event processing and offset tracking on the consumer side means events can be delivered once, with automatic retransmission de-duplication on failure. The Lightbend team says this provides reliable pub-sub between services while eliminating the burden of operating message broker infrastructure such as running an Apache Kafka or Apache Pulsar cluster.
The brokerless pub/sub also has event filtering with dynamic event filters on the producer or consumer side to prevent sending and processing unnecessary data to cut down on network use to minimize costs and use of hardware resources. The Lightbend team says this is particularly crucial in edge computing.
Another feature of the platform is active-active event sourcing. This makes use of CRDTs (conflict-free replicated data types, data structures that are replicated across multiple nodes in a network). It also uses the new brokerless pub/sub for efficient, low-latency replication to ensure eventual solid consistency of event-sourced entities across different data centers or Point-of-Presences at the edge.
The final feature of the platform is support for durable state queries. These can be used to search for data on multiple fields without additional reads, reducing the cost and time spent storing and interacting with duplicate data.
or email your comment to: firstname.lastname@example.org
|Last Updated ( Tuesday, 16 May 2023 )|