Defining the Cognitive Computing Continuum is only the first step. If the CCC is to become a mature research field, researchers also need evidence that allows them to understand what happens inside real distributed infrastructures. This is the OBSERVE stage of the newsletter’s progression. ENACT moves from conceptual foundations to measurable behaviour by making part of its own experimental evidence available to the wider community, turning project-generated telemetry into an open resource that other researchers can analyze, reproduce and reuse.
The publication “A telemetry dataset on resource utilisation and power consumption in the edge-cloud continuum” addresses an important limitation in current research. Real-world datasets that simultaneously capture resource utilization and energy consumption at both infrastructure level and application-workload level remain relatively scarce, particularly in heterogeneous cloud-edge environments. Many studies therefore depend on synthetic workloads, cloud-only traces or aggregated measurements that do not fully reflect the interactions occurring across the continuum.
ENACT’s dataset was generated through a distributed weather-forecasting scenario running on a Kubernetes cluster that combines a cloud-based machine and an edge device. The application is composed of five pods: two weather-data sources, two forecasting services deployed across different computing layers, and a long-term storage component. Using ENACT’s Telemetry Data Collector and Monitoring Engine, the experiment collected time-series information at both node and pod level, including CPU, memory and disk utilisation, network throughput and energy consumption. This combination creates a detailed picture of how infrastructure and workloads interact across heterogeneous resources.
A particularly useful aspect of the dataset is that measurements were collected under two distinct operating conditions: one with the application workloads running normally and a second baseline phase after the applications had been removed from the cluster. This makes it possible to examine the impact of application activity on resource usage and energy consumption rather than looking only at absolute values. For researchers interested in energy-aware computing, performance prediction or adaptive scheduling, that distinction is highly valuable because it helps separate workload effects from the underlying behaviour of the infrastructure.
The publication also makes clear that the dataset is intended for reuse. It can support the training or fine-tuning of time-series forecasting models, the benchmarking of anomaly-detection approaches, and the validation of scheduling algorithms, resource-optimization techniques and energy-aware strategies. This is also where the narrative begins to connect directly with the next article in the newsletter: the same type of empirical telemetry that helps us observe the continuum can also become an input for more intelligent mechanisms that decide how resources should be allocated.
From a dissemination perspective, this contribution captures the spirit of open knowledge particularly well. The scientific paper documents the methodology and value of the data, while the underlying dataset is made openly available through Zenodo. ENACT is therefore not only communicating what it has learned; it is providing other researchers with material that can be challenged, benchmarked and extended. Why it matters: this publication showcases ENACT’s contributions to open data by transforming ENACT’s internal experimental telemetry into a reusable community asset and creates an empirical bridge between understanding the CCC and developing the intelligent methods needed to optimize it.
Explore more in the reference publication: h https://doi.org/10.1016/j.dib.2026.112734
… and in the Open dataset on Zenodo: https://zenodo.org/records/18920397