2021/08/18 by Ajay Badita, Badita, Ajay, Rooji Jinan +5
Computer Science · #Advanced Data Storage Technologies #Caching and Content Delivery #Cloud Computing and Resource Management #Distributed #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Optimization and Search Problems #Parallel #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.2108.08199
openalex publication_date 2021/08/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We consider energy minimization for data-intensive applications run on large\nnumber of servers, for given performance guarantees. We consider a system,\nwhere each incoming application is sent to a set of servers, and is considered\nto be completed if a subset of them finish serving it. We consider a simple\ncase when each server core has two speed levels, where the higher speed can be\nachieved by higher power for each core independently. The core selects one of\nthe two speeds probabilistically for each incoming application request. We\nmodel arrival of application requests by a Poisson process, and random service\ntime at the server with independent exponential random variables. Our model and\nanalysis generalizes to today's state-of-the-art in CPU energy management where\neach core can independently select a speed level from a set of supported speeds\nand corresponding voltages. The performance metrics under consideration are the\nmean number of applications in the system and the average energy expenditure.\nWe first provide a tight approximation to study this previously intractable\nproblem and derive closed form approximate expressions for the performance\nmetrics when service times are exponentially distributed. Next, we study the\ntrade-off between the approximate mean number of applications and energy\nexpenditure in terms of the switching probability.\n