Ahmed Ali-Eldin

An adaptive hybrid elasticity controller for cloud infrastructures

Cloud elasticity is the ability of the cloud infrastructure to rapidly change the amount of resources allocated to a service in order to meet the actual varying demands on the service while enforcing SLAs. In this paper, we focus on horizontal elasticity, the ability of the infrastructure to add or remove virtual machines allocated to a service deployed in the cloud. We model a cloud service using queuing theory. Using that model we build two adaptive proactive controllers that estimate the future load on a service. We explore the different possible scenarios for deploying a proactive elasticity controller coupled with a reactive elasticity controller in the cloud. Using simulation with workload traces from the FIFA world-cup web servers, we show that a hybrid controller that incorporates a reactive controller for scale up coupled with our proactive controllers for scale down decisions reduces SLA violations by a factor of 2 to 10 compared to a regression based controller or a completely reactive controller.

Efficient provisioning of bursty scientific workloads on the cloud using adaptive elasticity control

Elasticity is the ability of a cloud infrastructure to dynamically change the amount of resources allocated to a running service as load changes. We build an autonomous elasticity controller that changes the number of virtual machines allocated to a service based on both monitored load changes and predictions of future load. The cloud infrastructure is modeled as a G/G/N queue. This model is used to construct a hybrid reactive-adaptive controller that quickly reacts to sudden load changes, prevents premature release of resources, takes into account the heterogeneity of the workload, and avoids oscillations. Using simulations with Web and cluster workload traces, we show that our proposed controller lowers the number of delayed requests by a factor of 70 for the Web traces and 3 for the cluster traces when compared to a reactive controller. Our controller also decreases the average number of queued requests by a factor of 3 for both traces, and reduces oscillations by a factor of 7 for the Web traces and 3 for the cluster traces. This comes at the expense of between 20% and 30% over-provisioning, as compared to a few percent for the reactive controller.

Self-management challenges for multi-cloud architectures

Addressing the management challenges for a multitude of distributed cloud architectures, we focus on the three complementary cloud management problems of predictive elasticity, admission control, and placement (or scheduling) of virtual machines. As these problems are intrinsically intertwined we also propose an approach to optimize the overall system behavior by policy-tuning for the tools handling each of them. Moreover, in order to facilitate the execution of some of the management decisions, we also propose new algorithms for live migration of virtual machines with very high workload and/or over low-bandwidth networks, using techniques such as caching, compression, and prioritization of memory pages.

The CACTOS Vision of Context-Aware Cloud Topology Optimization and Simulation

Recent advances in hardware development coupled with the rapid adoption and broad applicability of cloud computing have introduced widespread heterogeneity in data centers, significantly complicating the management of cloud applications and data center resources. This paper presents the CACTOS approach to cloud infrastructure automation and optimization, which addresses heterogeneity through a combination of in-depth analysis of application behavior with insights from commercial cloud providers. The aim of the approach is threefold: to model applications and data center resources, to simulate applications and resources for planning and operation, and to optimize application deployment and resource use in an autonomic manner. The approach is based on case studies from the areas of business analytics, enterprise applications, and scientific computing.

PEAS: A Performance Evaluation Framework for Auto-Scaling Strategies in Cloud Applications

Numerous auto-scaling strategies have been proposed in the past few years for improving various Quality of Service (QoS) indicators of cloud applications, for example, response time and throughput, by adapting the amount of resources assigned to the application to meet the workload demand. However, the evaluation of a proposed auto-scaler is usually achieved through experiments under specific conditions and seldom includes extensive testing to account for uncertainties in the workloads and unexpected behaviors of the system. These tests by no means can provide guarantees about the behavior of the system in general conditions. In this article, we present a Performance Evaluation framework for Auto-Scaling (PEAS) strategies in the presence of uncertainties. The evaluation is formulated as a chance constrained optimization problem, which is solved using scenario theory. The adoption of such a technique allows one to give probabilistic guarantees of the obtainable performance. Six different auto-scaling strategies have been selected from the literature for extensive test evaluation and compared using the proposed framework. We build a discrete event simulator and parameterize it based on real experiments. Using the simulator, each auto-scaler’s performance is evaluated using 796 distinct real workload traces from projects hosted on the Wikimedia foundations’ servers, and their performance is compared using PEAS. The evaluation is carried out using different performance metrics, highlighting the flexibility of the framework, while providing probabilistic bounds on the evaluation and the performance of the algorithms. Our results highlight the problem of generalizing the conclusions of the original published studies and show that based on the evaluation criteria, a controller can be shown to be better than other controllers.

Measuring Cloud Workload Burstiness

Workload burstiness and spikes are among the main reasons for service disruptions and decrease in the Quality-of-Service (QoS) of online services. They are hurdles that complicate autonomic resource management of datacenters. In this paper, we review the state-of-the-art in online identification of workload spikes and quantifying burstiness. The applicability of some of the proposed techniques is examined for Cloud systems where various workloads are co-hosted on the same platform. We discuss Sample Entropy (Samp En), a measure used in biomedical signal analysis, as a potential measure for burstiness. A modification to the original measure is introduced to make it more suitable for Cloud workloads.

A Survey on Modeling Energy Consumption of Cloud Applications: Deconstruction, State of the Art, and Trade-Off Debates

Given the complexity and heterogeneity in Cloud computing scenarios, the modeling approach has widely been employed to investigate and analyze the energy consumption of Cloud applications, by abstracting real-world objects and processes that are difficult to observe or understand directly. It is clear that the abstraction sacrifices, and usually does not need, the complete reflection of the reality to be modeled. Consequently, current energy consumption models vary in terms of purposes, assumptions, application characteristics and environmental conditions, with possible overlaps between different research works. Therefore, it would be necessary and valuable to reveal the state-of-the-art of the existing modeling efforts, so as to weave different models together to facilitate comprehending and further investigating application energy consumption in the Cloud domain. By systematically selecting, assessing, and synthesizing 76 relevant studies, we rationalized and organized over 30 energy consumption models with unified notations. To help investigate the existing models and facilitate future modeling work, we deconstructed the runtime execution and deployment environment of Cloud applications, and identified 18 environmental factors and 12 workload factors that would be influential on the energy consumption. In particular, there are complicated trade-offs and even debates when dealing with the combinational impacts of multiple factors.

Power-performance tradeoffs in data center servers: DVFS, CPU pinning, horizontal, and vertical scaling

Dynamic Voltage and Frequency Scaling (DVFS), CPU pinning, horizontal, and vertical scaling, are four techniques that have been proposed as actuators to control the performance and energy consumption on data center servers. This work investigates the utility of these four actuators, and quantifies the power-performance tradeoffs associated with them. Using replicas of the German Wikipedia running on our local testbed, we perform a set of experiments to quantify the influence of DVFS, vertical and horizontal scaling, and CPU pinning on end-to-end response time (average and tail), throughput, and power consumption with different workloads. Results of the experiments show that DVFS rarely reduces the power consumption of underloaded servers by more than 5%, but it can be used to limit the maximal power consumption of a saturated server by up to 20% (at a cost of performance degradation). CPU pinning reduces the power consumption of underloaded server (by up to 7%) at the cost of performance degradation, which can be limited by choosing an appropriate CPU pinning scheme. Horizontal and vertical scaling improves both the average and tail response time, but the improvement is not proportional to the amount of resources added. The load balancing strategy has a big impact on the tail response time of horizontally scaled applications. The impact of DVFS on the power consumption of underloaded servers is limited.vCPU consolidation reduces the power consumption at a cost of performance degradation.Combining VM scaling with consolidation of virtual CPUs improves energy efficiency.A load balancing strategy affects a tail latency of horizontally scaled applications.

Placement Matters: Replica Placement in Peer-Assisted Storage Clouds

Peer-assisted cloud storage systems use the unutilized resources of the clients subscribed to a storage cloud to offload the servers of the cloud. The provider distributes data replicas on the clients instead of replicating on the local infrastructure. These replicas allow the provider to provide a highly available, reliable and cheap service at a reduced cost. In this work we introduce Nile Store, a protocol for replication management in peer-assisted cloud storage. The protocol converts the replica placement problem into a linear task assignment problem. We design five utility functions to optimize placement taking into account the bandwidth, free storage and the size of data in need of replication on each peer. The problem is solved using a sub optimal greedy optimization algorithm. We show our simulation results using the different utilities under realistic network conditions. Our results show that using our approach offloads the cloud servers by about 90% compared to a random placement algorithm while consuming 98.5% less resources compared to a normal storage cloud.

Replica placement in peer-assisted clouds: an economic approach

We introduce NileStore, a replica placement algorithm based on an economical model for use in Peer-assisted cloud storage. The algorithm uses storage and bandwidth resources of peers to offload the cloud providers resources. We formulate the placement problem as a linear task assignment problem where the aim is to minimize time needed for file replicas to reach a certain desired threshold. Using simulation, We reduce the probability of a file being served from the providers servers by more than 97.5% under realistic network conditions.