The RACELab in the Computer Science Department at UCSB directed by Professors Chandra Krintz and Rich Wolski recently won a best paper award at the IEEE Cluster conference for their work on the next generation of cloud computing systems. The paper investigates how multiple "Big Data” analytics frameworks can be used more efficiently and effectively in Internet-of-Things (IoT) settings.
Key to their approach is moving of code (which is small) to where the data is collected or generated rather than moving vast amounts of data to the code, i.e. the Cloud, which is costly, wastes energy, and introduces network contention. The advance described in this paper describes a new resource allocator for Edge Clouds — private, resource-constrained systems that provide an intermediate computing infrastructure between IoT sensing devices and public cloud systems. The allocator facilitates low latency analytics that can be used to drive IoT actuation and control in a way that more fairly and with higher utilization, shares resources across multiple, concurrently executing analytics frameworks.
Justice: A Deadline-aware, Fair-share Resource Allocator for Implementing Multi-analytics
S. Dimopoulos, C. Krintz, and R. Wolski, Justice: A Deadline-aware, Fair-share Resource Allocator for Implementing Multi-analytics, IEEE Cluster, September, 2017.