Ops is an ambitious research arena that seeks to bring together Swedish industry and academia to solve state-of-the-art challenges in data-driven operational research.
It will develop an advanced data storage and processing infrastructure that will enable users to interactively store, annotate, process, and visualize large operational datasets, and – importantly – will provide a collaborative environment that will allow users to interact, share data and results, and set challenges and research questions. Ops aims to actively enable and encourage a significant increase in research collaborations between WASP community members, through a unique national resource.
The initial focus of WARA-Ops will be in two key areas:
We will collect and provide datasets and their associated research challenges, e.g., intrusion detection, denial-of-service attacks, etc. Examples of data include system/software log files, lists of access requests, etc.
We provide and generate datasets that can be used to drive research into next-generation autonomous management systems that can proactively predict future demand, detect, and react to changes in user/network/hardware behaviour, etc. Examples includes power reduction use metrics such as CPU/Disk/Network/Memory usage over time, and power consumption data.
The above research results are applicable across a wide set of application domains, including most online services, communication systems, automation services, energy management, financial and bank systems, online retail, and cloud/edge data centres. The application areas also include software build systems and CI/CD-pipelines.
With a strong focus on data storage and cataloging, combined with the provision of a scalable CPU-based environment for interactive data querying, processing, and visualisation, Ops is a natural complement to the Berzelius service (which focuses on GPU-intensive AI/ML workloads).
WARA Ops is a continuation of WARA Common. It will be professionally managed by Ericsson Research and significantly extends the current service. Additionally, Ops will include access to a live 5G network and full-scale cloud service, plus access to the monitoring data they generate. In addition to data generated within the Ops arena, the ambition is to also solicit data set through our partner network.
The expected research outputs of Ops are:
- data collection tools (including scalable storage, cataloging, and annotation services)
- methods for online processing (including auto-scaling resources for interactive data queries, filtering, and visualisation)
- a collaborative environment to facilitate interaction between researchers and partners (including ability to set research challenges on datasets, allow commenting and communication across researchers, etc.)
- and ultimately, self-learning anomaly detection mechanisms, root cause analysis strategies, and system theory for the control of large-scale systems.