CaSCON: Cache-Assisted Service Composition Orchestration Leveraging NDN

Contemporary applications in AR/VR, connected vehicles, Industry 4.0/5.0, and smart-buildings, have been recently gaining traction. These applications require orchestration of composite services in soft real-time to meet user Quality of Experience (QoE) performance. Furthermore, it is necessary to support seamless connectivity and computation at scale, despite uncertainties due to mobility and heterogeneity. Service composition orchestration solutions that are based on centralized orchestrators suffer from scalability challenges due to handling of data transmission between sub-services. The flexible forwarding and in-network caching capabilities of Named Data Networking (NDN) open the door to new distributed composition orchestration techniques. This paper develops Cache-Assisted Service Composition Orchestration leveraging NDN (CaSCON), aiming to reduce the latency of composite services through distributed orchestration. CaSCON allows services to forward results to succeeding services in the composite workflow, parallelizing execution based on data dependencies, and caching intermediate results of the pruned workflow within the network for potential future use. Experimental results through simulation, emulation, and hardware test-beds, show that these techniques are able to reduce communication and orchestration latency in the evaluated scenarios by 53% on average, and reduce transmitted packets by 52% on average, when compared to centralized orchestration.