Sara Casolari

Dynamic Load Management of Virtual Machines in Cloud Architectures

Cloud infrastructures must accommodate changing demands for different types of processing with heterogeneous workloads and time constraints. In a similar context, dynamic management of virtualized application environments is becoming very important to exploit computing resources, especially with recent virtualization capabilities that allow live sessions to be moved transparently between servers. This paper proposes novel management algorithms to decide about reallocations of virtual machines in a cloud context characterized by large numbers of hosts. The novel algorithms identify just the real critical instances and take decisions without recurring to typical thresholds. Moreover, they consider load trend behavior of the resources instead of instantaneous or average measures. Experimental results show that proposed algorithms are truly selective and robust even in variable contexts, thus reducing system instability and limit migrations when really necessary.

Models and framework for supporting runtime decisions in Web-based systems

Efficient management of distributed Web-based systems requires several mechanisms that decide on request dispatching, load balance, admission control, request redirection. The algorithms behind these mechanisms typically make fast decisions on the basis of the load conditions of the system resources. The architecture complexity and workloads characterizing most Web-based services make it extremely difficult to deduce a representative view of a resource load from collected measures that show extreme variability even at different time scales. Hence, any decision based on instantaneous or average views of the system load may lead to useless or even wrong actions. As an alternative, we propose a two-phase strategy that first aims to obtain a representative view of the load trend from measured system values and then applies this representation to support runtime decision systems. We consider two classical problems behind decisions: how to detect significant and nontransient load changes of a system resource and how to predict its future load behavior. The two-phase strategy is based on stochastic functions that are characterized by a computational complexity that is compatible with runtime decisions. We describe, test, and tune the two-phase strategy by considering as a first example a multitier Web-based system that is subject to different classes of realistic and synthetic workloads. Also, we integrate the proposed strategy into a framework that we validate by applying it to support runtime decisions in a cluster Web system and in a locally distributed Network Intrusion Detection System.

Dual time-scale distributed capacity allocation and load redirect algorithms for cloud systems

Resource management remains one of the main issues of cloud computing providers because system resources have to be continuously allocated to handle workload fluctuations while guaranteeing Service Level Agreements (SLA) to the end users. In this paper, we propose novel capacity allocation algorithms able to coordinate multiple distributed resource controllers operating in geographically distributed cloud sites. Capacity allocation solutions are integrated with a load redirection mechanism which, when necessary, distributes incoming requests among different sites. The overall goal is to minimize the costs of allocated resources in terms of virtual machines, while guaranteeing SLA constraints expressed as a threshold on the average response time. We propose a distributed solution which integrates workload prediction and distributed non-linear optimization techniques. Experiments show how the proposed solutions improve other heuristics proposed in literature without penalizing SLAs, and our results are close to the global optimum which can be obtained by an oracle with a perfect knowledge about the future offered load.

Load prediction models in web-based systems

Run-time management of modern Web-based services requires the integration of several algorithms and mechanisms for job dispatching, load sharing, admission control, overload detection. All these algorithms should take decisions on the basis of present and/or future load conditions of the system resources. In particular, we address the issue of predicting future resource loads under real-time constraints in the context of Internet-based systems. In this situation, it is extremely difficult to deduce a representative view of a system resource from collected raw measures that show very large variability even at different time scales. For this reason, we propose a two-step approach that first aims to get a representative view of the load trend from measured raw data, and then applies a load prediction algorithm to load trends. This approach is suitable to support different decision systems even for highly variable contexts and is characterized by a computational complexity that is compatible to run-time decisions. The proposed models are applied to a multi-tier Web-based system, but the results can be extended to other Internet-based contexts where the systems are characterized by similar workloads and resource behaviors.

Parameters of oxidative stress are present in the circulation of PXE patients

Pseudoxanthoma elasticum (PXE) is an inherited disorder characterized by calcification of elastic fibres leading to dermatological and vascular alterations associated to premature aged features and to life threatening clinical manifestations. The severity of the disease is independent from the type of mutation in the ABCC6 gene, and it has been suggested that local and/or systemic factors may contribute to the occurrence of clinical phenotype. The redox balance in the circulation of 27 PXE patients and of 50 healthy subjects of comparable age was evaluated by measuring the advanced oxidation protein products (AOPP), the lipid peroxidation derivatives (LOOH), the circulating total antioxidant status (TAS), the thiol content and the extracellular superoxide dismutase activity (EC-SOD). Patients were diagnosed by clinical, ultrastructural and molecular findings. Compared to control subjects, PXE patients exhibited significantly lower antioxidant potential, namely circulating TAS and free thiol groups, and higher levels of parameters of oxidative damage, as LOOH and of AOPP, and of circulating EC-SOD activity. Interestingly, the ratio between oxidant and antioxidant parameters was significantly altered in PXE patients and related to various score indices. This study demonstrates, for the first time, that several parameters of oxidative stress are modified in the blood of PXE patients and that the redox balance is significantly altered compared to control subjects of comparable age. Therefore, in PXE patients the circulating impaired redox balance may contribute to the occurrence of several clinical manifestations in PXE patients, and/or to the severity of disease, thus opening new perspectives for their management.

Flexible Distributed Capacity Allocation and Load Redirect Algorithms for Cloud Systems

In Cloud computing systems, resource management is one of the main issues. Indeed, in any time instant resources have to be allocated to handle effectively workload fluctuations, while providing Quality of Service (QoS) guarantees to the end users. In such systems, workload prediction-based autonomic computing techniques have been developed. In this paper we propose capacity allocation techniques able to coordinate multiple distributed resource controllers working in geographically distributed cloud sites. Furthermore, capacity allocation solutions are integrated with a load redirection mechanism which forwards incoming requests between different domains. The overall goal is to minimize the costs of the allocated virtual machine instances, while guaranteeing QoS constraints expressed as a threshold on the average response time. We compare multiple heuristics which integrate workload prediction and distributed non-linear optimization techniques. Experimental results show how our solutions significantly improve other heuristics proposed in the literature (5-35% on average), without introducing significant QoS violations.

Autonomic Request Management Algorithms for Geographically Distributed Internet-Based Systems

Supporting Web-based services through geographical distributed clusters of servers is a common solution to the increasing volume and variability of modern traffic. These architectures pose interesting challenges to request management strategies where the most important goal is not to achieve maximum performance, but to guarantee stable and robust results. In this paper, we propose novel request management algorithms that are based on autonomic principles that is, on loose collaboration among the closest nodes and no knowledge about the global system state. Experimental evaluation shows that our autonomic-enhanced algorithms can guarantee robust performance in a variety of settings and reduce standard deviations of the response times with respect to existing request management algorithms.

Short-term prediction models for server management in Internet-based contexts

Modern Internet applications run on top of complex system infrastructures where several runtime management algorithms have to guarantee high performance, scalability and availability. This paper aims to offer a support to runtime algorithms that must take decisions on the basis of historical and predicted load conditions of the internal system resources. We propose a new class of moving filtering techniques and of adaptive prediction models that are specifically designed to deal with runtime and short-term forecast of time series which originate from monitors of system resources of Internet-based servers. A large set of experiments confirm that the proposed models improve the prediction accuracy with respect to existing algorithms and they show stable results for different workload scenarios.

Dynamic load balancing for network intrusion detection systems based on distributed architectures

Increasing traffic and the necessity of stateful analyses impose strong computational requirements on network intrusion detection systems (NIDS), and motivate the need of distributed architectures with multiple sensors. In a context of high traffic with heavy tailed characteristics, static rules for dispatching traffic slices among distributed sensors cause severe imbalance. Hence, the distributed NIDS architecture must be combined with adequate mechanisms for dynamic load redistribution. In this paper, we propose and compare different policies for the activation/deactivation of the dynamic load balancer. In particular, we consider and compare single vs. double threshold schemes, and load representations based on resource measures vs. load aggregation models. Our experimental results show that the best combination of a double threshold scheme with a linear aggregation of resource measures is able to achieve a really satisfactory balance of the sensor loads together with a sensible reduction of the number of load balancer activations.

An adaptive model for online detection of relevant state changes in Internet-based systems

Modern Internet-based systems typically involve a large number of servers and applications and require real-time management strategies for cloning and migrating virtual machines, as well as re-distributing or re-mapping the underlying hardware. At the basis of most real-time management strategies there is the need to continuously evaluate system state behavior and to detect when a relevant state change is occurring. Modern Internet-based systems open new and interesting scenarios in the field of the research on the online state change detection models. In this paper, we propose an adaptive state change detection model that we demonstrate is suitable to analyze continuous streams of data coming from Internet-based systems characterized by high variability and non stationarity of the monitored resource measures that result in not-acceptable false alarm rates. Our model solves the limits of the traditional solutions while retaining their computational efficiency. The solution we present combines two key elements: an on-line wavelet model to denoise data streams and an adaptive detection rule. Experiments carried out using empirical and synthetic data sets confirm that the proposed method is able to signal all relevant state changes limiting the incorrect detections and to provide robust results even in non-stationary and highly variable contexts.