Kien Le

Reducing electricity cost through virtual machine placement in high performance computing clouds

In this paper, we first study the impact of load placement policies on cooling and maximum data center temperatures in cloud service providers that operate multiple geographically distributed data centers. Based on this study, we then propose dynamic load distribution policies that consider all electricity-related costs as well as transient cooling effects. Our evaluation studies the ability of different cooling strategies to handle load spikes, compares the behaviors of our dynamic cost-aware policies to cost-unaware and static policies, and explores the effects of many parameter settings. Among other interesting results, we demonstrate that (1) our policies can provide large cost savings, (2) load migration enables savings in many scenarios, and (3) all electricity-related costs must be considered at the same time for higher and consistent cost savings.

Capping the brown energy consumption of Internet services at low cost

The large amount of energy consumed by Internet services represents significant and fast-growing financial and environmental costs. Increasingly, services are exploring dynamic methods to minimize energy costs while respecting their service-level agreements (SLAs). Furthermore, it will soon be important for these services to manage their usage of “brown energy” (produced via carbon-intensive means) relative to renewable or “green” energy. This paper introduces a general, optimization-based framework for enabling multi-data-center services to manage their brown energy consumption and leverage green energy, while respecting their SLAs and minimizing energy costs. Based on the framework, we propose a policy for request distribution across the data centers. Our policy can be used to abide by caps on brown energy consumption, such as those that might arise from Kyoto-style carbon limits, from corporate pledges on carbon-neutrality, or from limits imposed on services to encourage brown energy conservation. We evaluate our framework and policy extensively through simulations and real experiments. Our results show how our policy allows a service to trade off consumption and cost. For example, using our policy, the service can reduce brown energy consumption by 24% for only a 10% increase in cost, while still abiding by SLAs.

Managing the cost, energy consumption, and carbon footprint of internet services

The large amount of energy consumed by Internet services represents significant and fast-growing financial and environmental costs. This paper introduces a general, optimization-based framework and several request distribution policies that enable multi-data-center services to manage their brown energy consumption and leverage green energy, while respecting their service-level agreements (SLAs) and minimizing energy cost. Our policies can be used to abide by caps on brown energy consumption that might arise from various scenarios such as government imposed Kyoto-style carbon limits. Extensive simulations and real experiments show that our policies allow a service to trade off consumption and cost. For example, using our policies, a service can reduce brown energy consumption by 24% for only a 10% increase in cost, while still abiding by SLAs.

Providing green SLAs in High Performance Computing clouds

Demand for clean products and services is increasing as society is becoming increasingly aware of climate change. In response, many enterprises are setting explicit sustainability goals and implementing initiatives to reduce carbon emissions. Quantification and disclosure of such goals and initiatives have become important marketing tools. As enterprises and individuals shift their workloads to the cloud, this drive toward quantification and disclosure will lead to demand for quantifiable green cloud services. Thus, we argue that cloud providers should offer a new class of green services, in addition to existing (energy-source-oblivious) services. This new class would provide clients with explicit service-level agreements (which we call Green SLAs) for the percentage of renewable energy used to run their workloads. In this paper, we first propose an approach for High Performance Computing cloud providers to offer such a Green SLA service. Specifically, each client job specifies a Green SLA, which is the minimum percentage of green energy that must be used to run the job. The provider earns a premium for meeting the Green SLA, but is penalized if it accepts the job but violates the Green SLA. We then propose (1) a power distribution and control infrastructure that uses a small amount of hardware to support Green SLAs, (2) an optimization-based framework for scheduling jobs and power sources that maximizes provider profits while respecting Green SLAs, and (3) two scheduling policies based on the framework. We evaluate our framework and policies extensively through simulations. Our main results show the tradeoffs between our policies, and their advantages over simpler greedy heuristics. We conclude that a Green SLA service that uses our policies would enable the provider to attract environmentally conscious clients, especially those who require strict guarantees on their use of green energy.

A cost-effective distributed file service with QoS guarantees

Large-scale, value-added Internet services composed of independent cooperating or competing services will soon become common place. Several groups have addressed the performance, communication, discovery, and description aspects of these services. However, little work has been done on effectively composing paid services and the quality-of-service (QoS) guarantees that they provide. We address these issues in the context of distributed file storage in this paper. In particular, we propose, implement, and evaluate a cost-effective, QoS-aware distributed file service comprising a front-end file service and back-end (third-party) storage services. Our front-end service uses mathematical modeling and optimization to provide performance and availability guarantees at low cost by carefully orchestrating the accesses to the back-end services. Experimental results from our prototype implementation validate our modeling and optimization. We conclude that our approach for providing QoS at low cost should be useful to future composite Internet services.