SLA-Tree: A Framework for Efficiently Supporting SLA-based Decisions in Cloud Computing Yun Chi Hyun Jin Moon Hakan Hacıgümü¸ s Junichi Tatemura NEC Laboratories America. 10080 North Wolfe Rd, SW3-350, Cupertino, CA 95014 {ychi,hjmoon,hakan,tatemura}@sv.nec-labs.com ABSTRACT As cloud computing becomes increasingly important in data- base systems, many new challenges and opportunities have arisen. One challenge is that in cloud computing, business profit plays a central role. Hence, it is very important for a cloud service provider to quickly make profit-oriented de- cisions. In this paper, we propose a novel data structure, called SLA-tree, to efficiently support profit-oriented deci- sion making. SLA-tree is built on two pieces of information: (1) a set of buffered queries waiting to be executed, which represents the scheduled events that will happen in the near future, and (2) a service level agreement (SLA) for each query, which indicates the different profits for the query for varying query response times. By constructing the SLA- tree, we efficiently support the answering of certain profit- oriented “what if” questions. Answers to these questions in turn can be applied to different profit-oriented decisions in cloud computing such as profit-aware scheduling, dispatch- ing, and capacity planning. Extensive experimental results based on both synthetic and real-world data demonstrate the effectiveness and efficiency of our SLA-tree framework. Categories and Subject Descriptors H.2 [Database Management]: Miscellaneous; E.1 [Data Structures]: Trees General Terms Algorithms, Management, Performance Keywords Cloud Computing, Service Level Agreement, SLA-tree, Schedul- ing, Dispatching, Capacity Planning 1. INTRODUCTION Accelerating adoption of the cloud computing introduces new challenges to traditional database systems [2, 6, 13]. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. EDBT 2011, March 22–24, 2011, Uppsala, Sweden. Copyright 2011 ACM 978-1-4503-0528-0/11/0003 ...$10.00 For example, a database service in the cloud may have to serve much more diverse clients than the traditional enter- prise databases usually do. One key observation is that cloud service providers have to optimize their profits while serv- ing diverse clients. Usually, the profit of a cloud service provider is determined by the services the provider offers to a variety of customers, governed by certain service level agreements (SLAs). As a consequence, for a cloud service provider, how to make intelligent profit-oriented decisions is a major challenge and a key to success. In this paper, we present a framework for efficiently supporting SLA-based, profit-oriented decisions in cloud computing. Figure 1: An example of cloud computing system. To further illustrate the challenges and to motivate our work, we use a real-life scenario. Assume a cloud service provider is hosting an online shopping site, where queries come from different users, as shown in Figure 1. When a query comes from a serious buyer (e.g., a user who is ready to check out, with high margin items in the shopping cart), the potential profit of answering the query can be high and the delay should be short; on the other hand, when the query is from a casual user (e.g., someone who just uses the shopping site’s tools to compare features of different products), the potential profit may be low and longer delay is tolerable. Yet another query may come from an internal employee who is collecting some data to make certain business decisions (e.g., whether to put certain products on sale), and in such a case a much longer delay is acceptable up to a certain threshold, after which a penalty may be incurred due to the failure to make the decision. In addition to different profit profiles, another observation from this example is that the workload in a cloud database system can be a mixture