Large-scale Cloud Infrastructures Research Articles

Carbon emission-aware job scheduling for Kubernetes deployments

Decreasing carbon emissions of data centers while guaranteeing Quality of Service (QoS) is one of the major challenges for efficient resource management of large-scale cloud infrastructures and societal sustainability. Previous works in the area of carbon reduction mostly focus on decreasing overall energy consumption, replacing energy sources with renewable ones, and migrating workloads to locations where lower emissions are expected. These measures do not consider the energy mix of the power used for the data center. In other words, all KWh of energy are considered the same from the point of view of emissions, which is rarely the case in practice. In this paper, we overcome this deficit by proposing a novel practical CO2-aware workload scheduling algorithm implemented in the Kubernetes orchestrator to shift non-critical jobs in time. The proposed algorithm predicts future CO2 emissions by using historical data of energy generation, selects time-shiftable jobs, and creates job schedules utilizing greedy sub-optimal CO2 decisions. The proposed algorithm is implemented using Kubernetes’ scheduler extender solution due to its ease of deployment with little overheads. The algorithm is evaluated with real-world workload traces and compared to the default Kubernetes scheduling implementation on several actual scenarios.

MAGNETIC: Multi-Agent Machine Learning-Based Approach for Energy Efficient Dynamic Consolidation in Data Centers

Improving the energy efficiency of data centers while guaranteeing Quality of Service (QoS), together with detecting performance variability of servers caused by either hardware or software failures, are two of the major challenges for efficient resource management of large-scale cloud infrastructures. Previous works in the area of dynamic Virtual Machine (VM) consolidation are mostly focused on addressing the energy challenge, but fall short in proposing comprehensive, scalable, and low-overhead approaches that jointly tackle energy efficiency and performance variability. Moreover, they usually assume over-simplistic power models, and fail to accurately consider all the delay and power costs associated with VM migration and host power mode transition. These assumptions are no longer valid in modern servers executing heterogeneous workloads and lead to unrealistic or inefficient results. In this paper, we propose a centralized-distributed low-overhead failure-aware dynamic VM consolidation strategy to minimize energy consumption in large-scale data centers. Our approach selects the most adequate power mode and frequency of each host during runtime using a distributed multi-agent Machine Learning (ML) based strategy, and migrates the VMs accordingly using a centralized heuristic. Our Multi-AGent machine learNing-based approach for Energy efficienT dynamIc Consolidation (MAGNETIC) is implemented in a modified version of the CloudSim simulator, and considers the energy and delay overheads associated with host power mode transition and VM migration, and is evaluated using power traces collected from various workloads running in real servers and resource utilization logs from cloud data center infrastructures. Results show how our strategy reduces data center energy consumption by up to 15 percent compared to other works in the state-of-the-art (SoA), guaranteeing the same QoS and reducing the number of VM migrations and host power mode transitions by up to 86 and 90 percent, respectively. Moreover, it shows better scalability than all other approaches, taking less than 0.7 percent time overhead to execute for a data center with 1,500 VMs. Finally, our solution is capable of detecting host performance variability due to failures, automatically migrating VMs from failing hosts and draining them from workload.

Big data security challenges and strategies

Big data, a recently popular term that refers to a massive collection of very large and complex data sets, is facing serious security and privacy challenges. Due to the typical characteristics of big data, namely velocity, volume and variety associated with large-scale cloud infrastructures and the Internet of Things (IoT), traditional security and privacy mechanisms are inadequate and unable to cope with the rapid data explosion in such a complex distributed computing environment. With big data analytics being widely used by businesses and government for decision making, security risk mitigation plays an important role in big data infrastructures worldwide. Traditional security mechanisms have failed to cope with the scalability, interoperability and adaptability of contemporary technologies that are required for big data. This paper takes an exploratory initial step using first principles to address this gap in literature. Firstly, we establish the current trends in big data comprehensively by identifying eleven Vs as important dimensions of big data, which form the contributing factors having an impact on the impending security problem. Next, we map the eleven Vs to the three phases of big data life cycle in order to unearth the major security and privacy challenges of big data. Finally, the paper provides four practical strategies adapted from contemporary technologies such as data provenance, encryption and access control, data mining and blockchain, identifying their associated real implementation examples. This work would pave way for future research investigations in this important big data security arena.

A framework for simulating large scale cloud infrastructures

Cloud infrastructures are continuously growing in size, since more cloud nodes are added to already existing hyper-scale infrastructures. These hyper-scale infrastructures are also becoming heterogeneous as different types of accelerators are added in order to increase performance per watt for certain types of applications and allow for various HPC workloads to migrate to Cloud environments. The introduction of diverse workloads that migrate in the Cloud along with increasing volume of incoming tasks results in phenomena of network congestion, underutilization and resource fragmentation. Simulators are used to analyze, study and possibly improve Cloud environments. However, existing Cloud simulation tools lack the ability to handle heterogeneous resources and tasks that span across multiple Cloud nodes. Moreover, they are mostly sequential and cannot scale to large numbers of Cloud nodes. Furthermore, they do not support over-commitment, which is a common practice in real-world Cloud environments. A framework for simulating large numbers of heterogeneous cloud nodes organized in Cells and executing large numbers of HPC tasks is proposed. The framework is inherently parallel and designed for hybrid distributed memory parallel systems, supporting CPU, memory and network over-commitment. The simulation framework is based on a time advancing loop, allowing dynamic change of the granularity of the simulator and minimizing memory requirements, since data related to the current time-step is stored. Moreover, a latency model for the currency of data in the Gateway Service and Broker is also supported. Implementation details along with discussions concerning the extensibility of the framework are given. Numerical results for simulating large number of heterogeneous resources and incoming tasks are also presented.

Introducing Fraudulent Energy Consumption in Cloud Infrastructures: A New Generation of Denial-of-Service Attacks

Currently, cloud computing is the target business environment for many enterprises and government organizations. However, despite the huge potential gains that can be achieved, security represents a fundamental issue, which prevents the massive cloud adoption in mission-critical Information Technology sectors. The most common security issues are amplified in the cloud environment since new complex features, with their inherent weaknesses, enter into the problem space, particularly those associated to multitenancy and elasticity. Thus, new threats, such as the energy-related denial-of-service attacks against large-scale cloud infrastructures, may involve not only the quality of the delivered services but also their operational costs in terms of energy bill. The longer is the time necessary to identify such attacks, the heavier is the impact on the overall energy consumption and, consequently, on the associated expenses. This work presents a detailed analysis of such new sophisticated menaces, by focusing on those that are specifically tailored to originate the worst-case energy demands by leveraging properly crafted low-rate traffic patterns in order to ensure stealth operations. We present some strategies exploiting the cloud flexibility in order to increase in a fraudulent way the overall energy consumption and analyze their impact within large-scale cloud infrastructures. This should help cloud providers in understanding such weaknesses and highlighting their root causes, as well as in providing some hints on how they can counter these subtle security issues.

Distributed decentralized collaborative monitoring architecture for cloud infrastructures

Cloud computing infrastructures are demanding an efficient monitoring mechanism to warranty the operational state of large-scale virtualized data centers and to provide mechanisms to improve the efficiency and stability of such infrastructures. Traditionally, centralized monitoring models (CMM) provide high performance and availability for the group of nodes in charge of monitoring tasks. However, the centralized nature of this architecture, easily leads to a single point of failure, bottlenecks in terms of performance and an unbalanced distributions of the monitoring workloads. These facts are not being suitable for large-scale cloud infrastructures. To tackle this concern, the main contribution of this paper is a distributed collaborative monitoring model (DCMM) for cloud computing infrastructures. DCMM provides self-organized capabilities based on mutual perception and balanced monitoring of each node. DCMM also provides rapid notification and recovery mechanisms under degraded conditions. In addition, an adaptive threshold control algorithm (ATCA) is proposed to dynamically adapt the sets of thresholds used for notification purposes in order to identify unnecessary duplicate information sent back to the monitoring tool. ATCA is based on historical monitoring records. Both DCMM and ATCA are described in detail in this contribution. Several empirical experiments have been done using OpenStack cloud infrastructure in order to validate our claims. Experimental results show that DCMM with ATCA can efficiently balance monitoring workload, reduce the workload of monitoring nodes, avoid a single point of failure, and reduce bottleneck problems whereas it is contributing to the achievement of real-time monitoring and data consistency within the monitoring architecture.

Evolutionary Computing Based on QoS Oriented Energy Efficient VM Consolidation Scheme for Large Scale Cloud Data Centers

Abstract The high pace and increase in cloud computing technology and associated applications, especially large scale data centres, have demanded energy efficient and Quality of Service (QoS) oriented computing platform. To meet these requirements, virtualization and Virtual Machine (VM) consolidation has emerged as an effective solution. The optimization in VM consolidation by means of efficient dynamic resource-utilization prediction, VM selection and placement can achieve optimal solution for energy efficient and QoS oriented cloud computing system. In this paper, an evolutionary computing algorithm called Adaptive Genetic Algorithm (A-GA) based VM consolidation approach has been developed. A-GA based placement policy and its implementation with different VM selection policies like Minimum Migration Time (MMT), Maximum Correlation (MC) and Random Selection (RS), along with different CPU utilization estimation approaches like Inter Quartile Range (IQR), Local Regression (LR), Local Robust Regression (LRR), static THReshold (THR) and Median Absolute Deviation (MAD) has revealed that A-GA based consolidation with MMT selection policy and combined IQR and LRR can enable optimal VM consolidation for large scale infrastructures. In addition, the proposed A-GA policy has exhibited better performance as compared to other meta-heuristics such as Ant Colony Optimization (ACO) and Best Fit Decreasing. The proposed consolidation system can be used for large scale cloud infrastructures where energy conservation, minimal Service Level Agreement (SLA) violation and QoS assurance is inevitable.

Intelligence based Defense System to Protect from Advanced Persistent Threat by means of Social Engineering on Social Cloud Platform

Advanced Persistent Threat is a specific design to survive until it reaches the target and creates backdoor for future Intrusion. The objective of this paper is, to design and implementation of a defense in depth system to protect social cloud against risks, threats and attacks. In this paper we discussed the policies and procedures of vulnerability management and intrusion detection system. We also focused on the steps involved in defense in depth system a layered architecture for prevention and protection from malicious users, masquerader and intruders. Where each layer restricts the intruder with deployed activities such as security monitoring, malicious site filtering, patches to remedy vulnerabilities, analysis of audit log files, Intrusion detection and prevention system and storage media management for protecting the cloud from advanced persistent threat. The different intrusion techniques using social engineering methods such as spear phishing or whale phishing are also discussed. A mathematical approach is given to secure the cloud computing using computational intelligence which is aimed to defense the attacks using Perron-Frobenius theorem and methods for public social network. In this paper we focused on the Defense in Depth model for a security layered strategy. We also discussed on the issues an audit communication or challenge to the cloud server data security. Defense System is the process of monitoring for and identifying unauthorized system access or manipulation and protecting the data from intruders. A clear description is given regarding the steps involved in the Advanced Persistent Attack and the measures or steps involved in Advanced Persistent Attack Defense System. The impact of security and privacy on cloud performance is calculated using cloud-sim tool, which supports simulating and modeling a large scale cloud infrastructure. An Experiment on intelligence based defense system using cloud analyst and simulator is given along with the result and conclusion.Keywords: Advanced Persistent Attack, Cloud Computing, Cloud Security, Cloud Server, Defense in Depth, Defense System, Vulnerability

An Adaptive Genetic Algorithm based Robust QoS Oriented Green Computing Scheme for VM Consolidation in Large Scale Cloud Infrastructures

Backgrounds: The high pace increase in cloud applications requires an optimal computing platform like, Virtual Machine (VM) Consolidationor virtualization to ensure optimal computational efficiency, energy consumption and minimal SLA violation. Methods: In this paper, an evolutionary computing approach called Adaptive Genetic Algorithm (A-GA) has been proposed for VM placement policy, to be used in VM consolidation. In the proposed model, the modified Robust Local Regression (LRR) and Inter-Quartile Range (IQR) schemes estimate the dynamic CPU utilization for overload detection, which is followed by Maximum Correlation (MC) and Minimum Migration Time (MMT) based VM selectionand A-GA based VM placement. Findings: The comparative performance analysis for the proposed system with Planet Lab cloud benchmark dataset has exhibited that the proposed model exhibits better results as compared to other heuristic approaches such as Best Fit Decreasing (BFD) algorithm and Ant Colony Optimization (ACO). The implementation of the proposed A-GA based consolidation with modified IQR and LRR, and MMT selection policyhas performed better in terms of energy efficiency and SLA violation as compared to the other heuristic approaches for placement such as Best Fit Decreasing (BFD) algorithm with conventional IQR, Local Regression (LR), robust local regression, Static Threshold (THR) and Median Absolute Deviation (MAD) based CPU utilization threshold estimation schemes. Furthermore, the proposed A-GA based scheme has outperformed Ant Colony Optimization (ACO) based consolidation scheme. The performance analysis with two distinct VM selection policies, MC and MMT has revealed that A-GA performs better with MMT selection policy and provides higher host shutdown, minimal VM migration and SLA violation, and minimal energy consumption. Applications: The proposed A-GAbased VM consolidation scheme can be significant for energy aware and QoS oriented virtualization application in large scale cloud infrastructures.

Recent Advances in Security and Privacy in Big Data

Big data has become an important topic in science, engineering, medicine, healthcare, finance, business and ultimately society itself. Big data refers to the massive amount of digital information stored or transmitted in computer systems. Approximately, 2.5 quintillion bytes of data are created every day. Almost 90% of data in the world today are created in the last two years alone. Security and privacy issues becomes more critical due to large volumes and variety, due to data hosted in large-scale cloud infrastructures, diversity of data sources and formats, streaming nature of data acquisition and high volume inter-cloud migration. In large-scale cloud infrastructures, a diversity of software platforms provides more opportunities to attackers. Traditional security mechanisms, which are usually invented for securing small-scale data, are inadequate. With a rapid growth of big data applications, it has become critical to introduce new security technology to accommodate the need of big data applications. The objective of this special issue is to capture the latest advances in this research field.

Distributed Computing Column 39

The relevance of Byzantine fault tolerance in the context of cloud computing has been questioned [3]. While arguments against Byzantine fault tolerance seemingly makes sense in the context of a single cloud, i.e., a large-scale cloud infrastructure that resides under control of a single, typically commercial provider, these arguments are less obvious in a broader context of the Intercloud, i.e., a cloud of multiple, independent clouds. In this paper, we start from commonly acknowledged issues that impede the adoption of Byzantine fault tolerance within a single cloud, and argue that many of these issues fade when Byzantine fault tolerance in the Intercloud is considered.

The byzantine empire in the intercloud

The relevance of Byzantine fault tolerance in the context of cloud computing has been questioned [3]. While arguments against Byzantine fault tolerance seemingly makes sense in the context of a single cloud, i.e., a large-scale cloud infrastructure that resides under control of a single, typically commercial provider, these arguments are less obvious in a broader context of the Intercloud, i.e., a cloud of multiple, independent clouds. In this paper, we start from commonly acknowledged issues that impede the adoption of Byzantine fault tolerance within a single cloud, and argue that many of these issues fade when Byzantine fault tolerance in the Intercloud is considered.