TR-2018-05
How to Increase the Value of Volatile Cloud Resources: Resource Management and Information Disclosure
Chaojie Zhang; Varun Gupta; Andrew A Chien. 14 April, 2018.
Communicated by Andrew Chien.
Abstract
Cloud providers sell unreliable or “volatile” resources that are unused by foreground (reserved/high priority) workloads. The value users can extract from these re- sources depends on (i) the volatile resource management algorithm, and (ii) the information provided to users about the volatile resources. We describe and evaluate four volatile resource management approaches (Random, FIFO, LIFO, LIFO-pools) using commercial cloud resource traces drawn from 608 Amazon EC2 instance pools. We also consider several information models (MTTR, limited statistics, Full distribution, and Oracle) that statistically characterize the resources for users.Our results show volatile resource management algorithms can increase user value by 30 to 45%. Slightly richer information models (90pctile) combined with LIFO and LIFO-pools volatile resource management increase user value by as much as 10x. Our results suggest that cloud providers should pay significant attention to what statistical information they provide to users. And, these results broadly characterize the vast majority (475 of 608) of instance pools. Finally, we provide a detailed drill- down showing how the volatile resource management algorithms affect resource interval durations, and thus potential user value. We further show how the information model shapes user targeting, success rate, and user value.
Original Document
The original document is available in PDF (uploaded 14 April, 2018 by Andrew Chien).