Akka

Overview

Edge apps extend edge computing by creating an application experience where services and data flow bidirectionally between IOT devices, far edge, near edge, and cloud systems. By integrating edge and cloud resources, edge apps deliver low-latency, high-performance solutions that optimize computation distribution and data flow.

What is an edge app?

Edge apps are daisy-chained services that span from edge devices through intermediate processing layers and into the cloud, with data flowing bi-directionally.

Edge applications have different logic and data services that operate on each node of the chained application, often limited by the resources available within the execution environment. Edge apps aggregate, stream, and share data with nodes that are upstream and downstream from each other. They need adaptive resource utilization, resilience and autonomy operating in unreliable networks, and context awareness.

Key properties of edge apps

Edge applications depend upon compute elasticity, location transparency, and network-based context awareness to operate different services as a whole system.

Daisy-chained services

Edge apps have a bi-directional and network-reliable flow of data and analytics to upstream and downstream services that are part of the app.

Context awareness

Edge apps must adapt behavior based on local conditions, available resources, and user proximity to optimize performance and user experience.

Distributed processing

Services must operate consistently despite being deployed across a diverse set of computational resources. For example, the computing environment on a far-edge node is not identical to the cloud.

Bi-directional data flow

Efficient data movement between edge and cloud in both directions enables adaptive behavior and real-time updates. Data analysis happens near the source and complex analysis happens near the cloud.

Adaptive resource utilization

Edge apps must dynamically adjust their resource usage and processing location. This adaptability optimizes performance and efficiency across varying computational environments.

Resilience and autonomy

Applications should maintain functionality during network instability, operate independently when disconnected from the cloud, and synchronize state when rejoining after a disconnect.

How Akka enables edge apps

Akka enables the development and operation of elastic, distributed applications whose services process data and logic across the range of node types in edge computing: IOT devices, far edge, near edge, and cloud. Akka has a common development model for creating edge services wired to communicate and operate with one another.

Reliable brokerless messaging

Akka enables asynchronous, reliable gRPC messaging over unreliable networks to enable service communication and data flows between edge app nodes.

Consuming and Producing

Consume from Event Sourced Entity

Consume from Key Value Entity

Data synchronization & replication

Akka enables the creation of stateful application services with entities whose state changes can be synchronized or replicated with upstream and downstream peers.

Fully Replicated Entities

Services on IOT devices

Akka services can be built in Rust and compiled into min-resource (25MB RAM) binaries for deployment onto IoT devices. Through its event-sourcing capabilities, Akka manages local storage on IoT devices, allowing them to maintain critical data locally and synchronize with other nodes.

Akka Edge Rust

Edge-to-cloud programming model

The programming model for building services and data systems is identical whether the service is deployed in the cloud or on an edge node.

Akka Innovations

Horizontal scale out to millions of nodes

Akka services self-cluster and can execute horizontal scale out across millions of nodes. Data within nodes is sharded and re-partitioned as the system scales. Traffic is automatically steered to the node best suited to handling the request.

Physical deployment

Akka Distributed Cluster

Architectural Overview

Distributed data pipelines

Build efficient, real-time, continuous data processing services that chain their imports and exports across the network. Replicate results and perform result aggregations propagated to upstream and downstream services.

Active-Active Entities