Cosmin Rusu

A performance-conserving approach for reducing peak power consumption in server systems

The combination of increasing component power consumption, a desire for denser systems, and the required performance growth in the face of technology-scaling issues are posing enormous challenges for powering and cooling of server systems. The challenges are directly linked to the peak power consumption of servers.Our solution, Power Shifting, reduces the peak power consumption of servers minimizing the impact on performance. We reduce peak power consumption by using workload-guided dynamic allocation of power among components incorporating real-time performance feedback, activity-related power estimation techniques, and performance-sensitive activity-regulation mechanisms to enforce power budgets.We apply our techniques to a computer system with a single processor and memory. Power shifting adds a system power manager with a dynamic, global view of the systems power consumption to continuously re-budget the available power amongst the two components. Our contributions include:• Demonstration of the greater effectiveness of dynamic power allocation over static budgeting,• Evaluation of different power shifting policies,• Analysis of system and workload factors critical to successful power shifting, and• Proposal of performance-sensitive power budget enforcement mechanisms that ensure system reliability.

Energy-Efficient Real-Time Heterogeneous Server Clusters

With increasing costs of energy consumption and cooling, power management in server clusters has become an increasingly important design issue. Current clusters for real-time applications are designed to handle peak loads, where all servers are fully utilized. In practice, peak load conditions rarely happen and clusters are most of the time underutilized. This creates the opportunity for using slower frequencies, and thus smaller energy consumption, with little or no impact on the Quality of Service (QoS), for example, performance and timeliness. In this work we present a cluster-wide QoS-aware technique that dynamically reconfigures the cluster to reduce energy consumption during periods of reduced load. Moreover, we also investigate the effects of local QoS-aware power management using Dynamic Voltage Scaling (DVS). Since most real-world clusters consist of machines of different kind (in terms of both performance and energy consumption) we focus on heterogeneous clusters. For validation, we describe and evaluate an implementation of the proposed scheme using the Apache Webserver in a small realistic cluster. Our experimental results show that using our scheme it is possible to save up to 45% of the total energy consumed by the servers, maintaining average response times within the specified deadlines and number of dropped requests within the required amount.

Multiversion scheduling in rechargeable energy-aware real-time systems

In the context of battery-powered real-time systems three constraints need to be addressed: energy; deadlines; and task rewards. Many future real-time systems will count on different software versions, each with different rewards, time and energy requirements, to achieve a variety of QoS-aware tradeoffs. We propose a solution that allows the device to run the most valuable task versions while still meeting all deadlines and without depleting the energy. Assuming that the battery is rechargeable, we also propose: (a) a static solution that maximizes the system value assuming a worst-case scenario (i.e., worst-case task execution times); and (b) a dynamic scheme that takes advantage of the extra energy in the system when worst-case scenarios do not happen. Three dynamic policies are shown to make better use of the recharging energy while improving the system value.

Maximizing rewards for real-time applications with energy constraints

New technologies have brought about a proliferation of embedded systems, which vary from control systems to sensor networks to personal digital assistants. Many of the portable embedded devices run several applications, which typically have three constraints that need to be addressed: energy, deadline, and reward. However, many of these portable devices do not have powerful enough CPUs and batteries to run all applications within their deadlines. An optimal scheme would allow the device to run the most applications, each using the most amount of CPU cycles possible, without depleting the energy source while still meeting all deadlines. In this paper we propose a solution to this problem; to our knowledge, this is the first solution that combines the three constraints mentioned above. We devise two algorithms, an optimal algorithm for homogeneous applications (with respect to power consumption) and a heuristic iterative algorithm that can also accommodate heterogeneous applications (i.e., those with different power consumption functions). We show by simulation that our iterative algorithm is fast and within 1% of the optimal.

Energy-efficient policies for embedded clusters

Power conservation has become a key design issue for many systems, including clusters deployed for embedded systems, where power availability ultimately determines system lifetime. These clusters execute a high rate of requests of highly-variable length, such as in satellite-based multiprocessor systems. The goal of power management in such systems is to minimize the aggregate energy consumption of the whole cluster while ensuring timely responses to requests. In the past, dynamic voltage scaling (DVS) and on/off schemes have been studied under the assumptions of continuously tunable processor frequencies and perfect load-balancing. In this work, we focus on the more realistic case of discrete processor frequencies and propose a new policy that adjusts the number of active nodes based on the system load, not system frequency. We also design a threshold scheme which prevents the system from reacting to short-lived temporary workload changes in the presence of unstable incoming workload. Simulation and implementation results on real hardware show that our policy is very effective in reducing the overall power consumption of clusters executing embedded applications.

Maximizing the system value while satisfying time and energy constraints

Embedded devices designed for various real-time applications typically have three constraints that must be addressed: energy, deadlines, and reward. These constraints play important roles in the next generation of embedded systems, since they provide users with a variety of quality-of-service (QoS) tradeoffs. We propose a QoS model in which applications may have several versions, each with different time and energy requirements, while providing different levels of accuracy (reward). An optimal scheme would allow the device to run the most critical and valuable versions of applications without depleting the energy source, while still meeting all deadlines. A solution is presented for frame-based and periodic task sets. Three algorithms are devised that closely approximate the optimal solution while taking only a fraction of the runtime of an optimal solution.

Energy-efficient policies for request-driven soft real-time systems

Computing systems, ranging from small battery-operated embedded systems to more complex general purpose systems, are designed to satisfy various computation demands in some acceptable time. In doing so, the system is responsible for scheduling jobs/requests in a dynamic fashion. In addition, with power consumption recently becoming a critical issue, most systems are also responsible for their own power management. In some rare cases, the exact arrival time and execution time of jobs/requests is known, leading to precise scheduling algorithms and power management schemes. However, more often than not, there is no a-priori knowledge of the workload. This work evaluates dynamic voltage scaling (DVS) policies for power management in systems with unpredictable workloads. A clear winner is identified, a policy that reduces the energy consumption one order of magnitude compared to no power management and up to 40% (in real-life traces) and 50% (in synthetic workloads) compared to the second-best evaluated scheme.

Integrated CPU and l2 cache voltage scaling using machine learning

Embedded systems serve an emerging and diverse set of applications. As a result, more computational and storage capabilities are added to accommodate ever more demanding applications. Unfortunately, adding more resources typically comes on the expense of higher energy costs. New chip design with Multiple Clock Domains (MCD) opens the opportunity for fine-grain power management within theprocessor chip. When used with dynamic voltage scaling (DVS), we can control the voltage and power of each domain independently. A significant power and energy improvement has been shown when using MCD design in comparison to managing a single voltage domain for the whole chip, as in traditional chips with global DVS. In this paper, we propose PACSL a Power-Aware Compiler-based approach using Supervised Learning. PACSL automatically derives an integrated CPU-core and on-chip L2 cache DVS policy tailored to a specific system and workload. Our approach uses supervised machine learning to discover a policy, which relies on monitoring a few performance counters. We present our approach detailing the role of a compiler in constructing a custom power management policy. We also discuss some implementation issues associated with our technique. We show that PACSL improves on traditional power management techniques that are used in general MCD chips. Our technique saves 22% on average (up to 46%) in energy-delay product over a DVS technique that applies independent DVS decisions in each domain. Compared to no-power management, our technique improves energy-delay product by 26% on average (up to 64%).