Dynamic load balancing in parallel KD-tree k-means

One among the most influential and popular data mining methods is the k-Means algorithm for cluster analysis. Techniques for improving the efficiency of k-Means have been largely explored in two main directions. The amount of computation can be significantly reduced by adopting geometrical constraints and an efficient data structure, notably a multidimensional binary search tree (KD-Tree). These techniques allow to reduce the number of distance computations the algorithm performs at each iteration. A second direction is parallel processing, where data and computation loads are distributed over many processing nodes. However, little work has been done to provide a parallel formulation of the efficient sequential techniques based on KD-Trees. Such approaches are expected to have an irregular distribution of computation load and can suffer from load imbalance. This issue has so far limited the adoption of these efficient k-Means variants in parallel computing environments. In this work, we provide a parallel formulation of the KD-Tree based k-Means algorithm for distributed memory systems and address its load balancing issue. Three solutions have been developed and tested. Two approaches are based on a static partitioning of the data set and a third solution incorporates a dynamic load balancing policy.

Load balancing solution and evaluation of F5 content switch equipment

The Thesis focused on hardware based Load balancing solution of web traffic through a load balancer F5 content switch. In this project, the implemented scenario for distributing HTTPtraffic load is based on different CPU usages (processing speed) of multiple member servers.Two widely used load balancing algorithms Round Robin (RR) and Ratio model (weighted Round Robin) are implemented through F5 load balancer. For evaluating the performance of F5 content switch, some experimental tests has been taken on implemented scenarios using RR and Ratio model load balancing algorithms. The performance is examined in terms of throughput (bits/sec) and Response time of member servers in a load balancing pool. From these experiments we have observed that Ratio Model load balancing algorithm is most suitable in the environment of load balancing servers with different CPU usages as it allows assigning the weight according to CPU usage both in static and dynamic load balancing of servers.

A study of adaptive load balancing algorithms for distributed systems

With the advent of distributed computer systems with a largely transparent user interface, new questions have arisen regarding the management of such an environment by an operating system. One fertile area of research is that of load balancing, which attempts to improve system performance by redistributing the workload submitted to the system by the users. Early work in this field concentrated on static placement of computational objects to improve performance, given prior knowledge of process behaviour. More recently this has evolved into studying dynamic load balancing with process migration, thus allowing the system to adapt to varying loads. In this thesis, we describe a simulated system which facilitates experimentation with various load balancing algorithms. The system runs under UNIX and provides functions for user processes to communicate through software ports; processes reside on simulated homogeneous processors, connected by a user-specified topology, and a mechanism is included to allow migration of a process from one processor to another. We present the results of a study of adaptive load balancing algorithms, conducted using the aforementioned simulated system, under varying conditions; these results show the relative merits of different approaches to the load balancing problem, and we analyse the trade-offs between them. Following from this study, we present further novel modifications to suggested algorithms, and show their effects on system performance.

Load-balancing for parallel adaptive unstructured grids

A parallel method for the dynamic partitioning of unstructured meshes is outlined. The method includes diffusive load-balancing techniques and an iterative optimisation technique known as relative gain optimisationwhich both balances theworkload and attempts to minimise the interprocessor communications overhead. It can also optionally include amultilevel strategy. Experiments on a series of adaptively refined meshes indicate that the algorithmprovides partitions of an equivalent or higher quality to static partitioners (which do not reuse the existing partition) and much more rapidly. Perhaps more importantly, the algorithm results in only a small fraction of the amount of data migration compared to the static partitioners.

Parallel dynamic load-balancing for adaptive unstructured meshes

This chapter describes a parallel optimization technique that incorporates a distributed load-balancing algorithm and provides an extremely fast solution to the problem of load-balancing adaptive unstructured meshes. Moreover, a parallel graph contraction technique can be employed to enhance the partition quality and the resulting strategy outperforms or matches results from existing state-of-the-art static mesh partitioning algorithms. The strategy can also be applied to static partitioning problems. Dynamic procedures have been found to be much faster than static techniques, to provide partitions of similar or higher quality and, in comparison, involve the migration of a fraction of the data. The method employs a new iterative optimization technique that balances the workload and attempts to minimize the interprocessor communications overhead. Experiments on a series of adaptively refined meshes indicate that the algorithm provides partitions of an equivalent or higher quality to static partitioners (which do not reuse the existing partition) and much more quickly. The dynamic evolution of load has three major influences on possible partitioning techniques; cost, reuse, and parallelism. The unstructured mesh may be modified every few time-steps and so the load-balancing must have a low cost relative to that of the solution algorithm in between remeshing.

Dynamic load-balancing for parallel adaptive unstructured meshes

A parallel method for dynamic partitioning of unstructured meshes is described. The method employs a new iterative optimisation technique which both balances the workload and attempts to minimise the interprocessor communications overhead. Experiments on a series of adaptively refined meshes indicate that the algorithm provides partitions of an equivalent or higher quality to static partitioners (which do not reuse the existing partition) and much more quickly. Perhaps more importantly, the algorithm results in only a small fraction of the amount of data migration compared to the static partitioners.

Dynamic load balancing of distributed memory parallel computational mechanics using unstructured meshes for multi-physical modelling

As the complexity of parallel applications increase, the performance limitations resulting from computational load imbalance become dominant. Mapping the problem space to the processors in a parallel machine in a manner that balances the workload of each processors will typically reduce the run-time. In many cases the computation time required for a given calculation cannot be predetermined even at run-time and so static partition of the problem returns poor performance. For problems in which the computational load across the discretisation is dynamic and inhomogeneous, for example multi-physics problems involving fluid and solid mechanics with phase changes, the workload for a static subdomain will change over the course of a computation and cannot be estimated beforehand. For such applications the mapping of loads to process is required to change dynamically, at run-time in order to maintain reasonable efficiency. The issue of dynamic load balancing are examined in the context of PHYSICA, a three dimensional unstructured mesh multi-physics continuum mechanics computational modelling code.

Mesh partitioning and load balancing for distributed memory parallel systems

A method is outlined for optimising graph partitions which arise in mapping unstructured mesh calculations to parallel computers. The method employs a relative gain iterative technique to both evenly balance the workload and minimise the number and volume of interprocessor communications. A parallel graph reduction technique is also briefly described and can be used to give a global perspective to the optimisation. The algorithms work efficiently in parallel as well as sequentially and when combined with a fast direct partitioning technique (such as the Greedy algorithm) to give an initial partition, the resulting two-stage process proves itself to be both a powerful and flexible solution to the static graph-partitioning problem. Experiments indicate that the resulting parallel code can provide high quality partitions, independent of the initial partition, within a few seconds. The algorithms can also be used for dynamic load-balancing, reusing existing partitions and in this case the procedures are much faster than static techniques, provide partitions of similar or higher quality and, in comparison, involve the migration of a fraction of the data.

Evaluation of PostgreSQL replication and load balancing implementations

In this final project the high availability options for PostgreSQL database management system were explored and evaluated. The primary objective of the project was to find a reliable replication system and implement it to a production environment. The secondary objective was to explore different load balancing methods and compare their performance. The potential replication methods were thoroughly examined, and the most promising was implemented to a database system gathering weather information in Lithuania. The different load balancing methods were tested performance wise with different load scenarios and the results were analysed. As a result for this project a functioning PostgreSQL database replication system was built to the Lithuanian Hydrometeorological Service's headquarters, and definite guidelines for future load balancing needs were produced. This study includes the actual implementation of a replication system to a demanding production environment, but only guidelines for building a load balancing system to the same production environment.

Load balancing in P2P smartphone based distributed IoT systems

With the new age of Internet of Things (IoT), object of everyday such as mobile smart devices start to be equipped with cheap sensors and low energy wireless communication capability. Nowadays mobile smart devices (phones, tablets) have become an ubiquitous device with everyone having access to at least one device. There is an opportunity to build innovative applications and services by exploiting these devices’ untapped rechargeable energy, sensing and processing capabilities. In this thesis, we propose, develop, implement and evaluate LoadIoT a peer-to-peer load balancing scheme that can distribute tasks among plethora of mobile smart devices in the IoT world. We develop and demonstrate an android-based proof of concept load-balancing application. We also present a model of the system which is used to validate the efficiency of the load balancing approach under varying application scenarios. Load balancing concepts can be apply to IoT scenario linked to smart devices. It is able to reduce the traffic send to the Cloud and the energy consumption of the devices. The data acquired from the experimental outcomes enable us to determine the feasibility and cost-effectiveness of a load balanced P2P smart phone-based applications.

Dynamic load balancing for the distributed mining of molecular structures

In molecular biology, it is often desirable to find common properties in large numbers of drug candidates. One family of methods stems from the data mining community, where algorithms to find frequent graphs have received increasing attention over the past years. However, the computational complexity of the underlying problem and the large amount of data to be explored essentially render sequential algorithms useless. In this paper, we present a distributed approach to the frequent subgraph mining problem to discover interesting patterns in molecular compounds. This problem is characterized by a highly irregular search tree, whereby no reliable workload prediction is available. We describe the three main aspects of the proposed distributed algorithm, namely, a dynamic partitioning of the search space, a distribution process based on a peer-to-peer communication framework, and a novel receiverinitiated load balancing algorithm. The effectiveness of the distributed method has been evaluated on the well-known National Cancer Institute’s HIV-screening data set, where we were able to show close-to linear speedup in a network of workstations. The proposed approach also allows for dynamic resource aggregation in a non dedicated computational environment. These features make it suitable for large-scale, multi-domain, heterogeneous environments, such as computational grids.

Decentralized load balancing for highly irregular search problems

In this paper, we present a distributed computing framework for problems characterized by a highly irregular search tree, whereby no reliable workload prediction is available. The framework is based on a peer-to-peer computing environment and dynamic load balancing. The system allows for dynamic resource aggregation, does not depend on any specific meta-computing middleware and is suitable for large-scale, multi-domain, heterogeneous environments, such as computational Grids. Dynamic load balancing policies based on global statistics are known to provide optimal load balancing performance, while randomized techniques provide high scalability. The proposed method combines both advantages and adopts distributed job-pools and a randomized polling technique. The framework has been successfully adopted in a parallel search algorithm for subgraph mining and evaluated on a molecular compounds dataset. The parallel application has shown good calability and close-to linear speedup in a distributed network of workstations.

A load-balancing spare capacity reallocation approach in service-rich SONET metro mesh networks

The next-generation SONET metro network is evolving into a service-rich infrastructure. At the edge of such a network, multi-service provisioning platforms (MSPPs) provide efficient data mapping enabled by Generic Framing Procedure (GFP) and Virtual Concatenation (VC). The core of the network tends to be a meshed architecture equipped with Multi-Service Switches (MSSs). In the context of these emerging technologies, we propose a load-balancing spare capacity reallocation approach to improve network utilization in the next-generation SONET metro networks. Using our approach, carriers can postpone network upgrades, resulting in increased revenue with reduced capital expenditures (CAPEX). For the first time, we consider the spare capacity reallocation problem from a capacity upgrade and network planning perspective. Our approach can operate in the context of shared-path protection (with backup multiplexing) because it reallocates spare capacity without disrupting working services. Unlike previous spare capacity reallocation approaches which aim at minimizing total spare capacity, our load-balancing approach minimizes the network load vector (NLV), which is a novel metric that reflects the network load distribution. Because NLV takes into consideration both uniform and non-uniform link capacity distribution, our approach can benefit both uniform and non-uniform networks. We develop a greedy loadbalancing spare capacity reallocation (GLB-SCR) heuristic algorithm to implement this approach. Our experimental results show that GLB-SCR outperforms a previously proposed algorithm (SSR) in terms of established connection capacity and total network capacity in both uniform and non-uniform networks.

A Load-Balancing Shared-Protection-Path Reconfiguration Approach in WDM Wavelength-Routed Networks

Wavelength-routed networks (WRN) are very promising candidates for next-generation Internet and telecommunication backbones. In such a network, optical-layer protection is of paramount importance due to the risk of losing large amounts of data under a failure. To protect the network against this risk, service providers usually provide a pair of risk-independent working and protection paths for each optical connection. However, the investment made for the optical-layer protection increases network cost. To reduce the capital expenditure, service providers need to efficiently utilize their network resources. Among all the existing approaches, shared-path protection has proven to be practical and cost-efficient [1]. In shared-path protection, several protection paths can share a wavelength on a fiber link if their working paths are risk-independent. In real-world networks, provisioning is usually implemented without the knowledge of future network resource utilization status. As the network changes with the addition and deletion of connections, the network utilization will become sub-optimal. Reconfiguration, which is referred to as the method of re-provisioning the existing connections, is an attractive solution to fill in the gap between the current network utilization and its optimal value [2]. In this paper, we propose a new shared-protection-path reconfiguration approach. Unlike some of previous reconfiguration approaches that alter the working paths, our approach only changes protection paths, and hence does not interfere with the ongoing services on the working paths, and is therefore risk-free. Previous studies have verified the benefits arising from the reconfiguration of existing connections [2] [3] [4]. Most of them are aimed at minimizing the total used wavelength-links or ports. However, this objective does not directly relate to cost saving because minimizing the total network resource consumption does not necessarily maximize the capability of accommodating future connections. As a result, service providers may still need to pay for early network upgrades. Alternatively, our proposed shared-protection-path reconfiguration approach is based on a load-balancing objective, which minimizes the network load distribution vector (LDV, see Section 2). This new objective is designed to postpone network upgrades, thus bringing extra cost savings to service providers. In other words, by using the new objective, service providers can establish as many connections as possible before network upgrades, resulting in increased revenue. We develop a heuristic load-balancing (LB) reconfiguration approach based on this new objective and compare its performance with an approach previously introduced in [2] and [4], whose objective is minimizing the total network resource consumption.

Tecniche Cloud Avanzate per Adaptive Hadoop Load Balancing tramite Sahara

Ogni giorno vengono generati grandi moli di dati attraverso sorgenti diverse. Questi dati, chiamati Big Data, sono attualmente oggetto di forte interesse nel settore IT (Information Technology). I processi digitalizzati, le interazioni sui social media, i sensori ed i sistemi mobili, che utilizziamo quotidianamente, sono solo un piccolo sottoinsieme di tutte le fonti che contribuiscono alla produzione di questi dati. Per poter analizzare ed estrarre informazioni da questi grandi volumi di dati, tante sono le tecnologie che sono state sviluppate. Molte di queste sfruttano approcci distribuiti e paralleli. Una delle tecnologie che ha avuto maggior successo nel processamento dei Big Data, e Apache Hadoop. Il Cloud Computing, in particolare le soluzioni che seguono il modello IaaS (Infrastructure as a Service), forniscono un valido strumento all'approvvigionamento di risorse in maniera semplice e veloce. Per questo motivo, in questa proposta, viene utilizzato OpenStack come piattaforma IaaS. Grazie all'integrazione delle tecnologie OpenStack e Hadoop, attraverso Sahara, si riesce a sfruttare le potenzialita offerte da un ambiente cloud per migliorare le prestazioni dell'elaborazione distribuita e parallela. Lo scopo di questo lavoro e ottenere una miglior distribuzione delle risorse utilizzate nel sistema cloud con obiettivi di load balancing. Per raggiungere questi obiettivi, si sono rese necessarie modifiche sia al framework Hadoop che al progetto Sahara.