Crowdsourced Resource-Sizing of Virtual Appliances

• Main Authors: Delul, Pinar Yanardag, Griffith, Rean, Holler, Anne, Shankari, K., Xiaoyun Zhu, Soundararajan, Ravi, Jagadeeshwaran, Adarsh, Padala, Pradeep
• Format: Conference Proceeding
• Language: English
• Subjects: Feature extraction; Home appliances; Linux; Measurement; Mutual information; Predictive models; Virtual machining
• ISSN: 2159-6182; 2159-6190
• DOI: 10.1109/CLOUD.2014.111

Using a population of VMware Virtual Center Virtual Appliances (VCVA) and their respective workloads we de- scribe techniques for constructing a model of their resource consumption and performance, specially memory requirements, and average operation-latency by mining logs of application (VCVA) performance. We use our model to provide sizing recommendations for the virtual appliance and identify features that can be used to provide rough estimates of expected memory consumption. We show results of better than 70% prediction accuracy (recall) for predicting Physical Memory Usage and better than 80% prediction accuracy (recall) for predicting the average latency of work- load operations. We describe modeling techniques from statistical machine learning that are amenable to representing complex, non-linear systems. Further, via the choice of techniques, we present an approach for reasoning about the limitations of our model, i.e., identifying when (and why) our model is expected to perform well and poorly.