Search results for "Load balancing"
showing 10 items of 49 documents
Accelerating Application Migration in HPC
2016
It is predicted that the number of cores per node will rapidly increase with the upcoming era of exascale supercomputers. As a result, multiple applications will have to share one node and compete for the (often scarce) resources available on this node. Furthermore, the growing number of hardware components causes a decrease in the mean time between failures. Application migration between nodes has been proposed as a tool to mitigate these two problems: Bottlenecks due to resource sharing can be addressed by load balancing schemes which migrate applications; and hardware errors can often be tolerated by the system if faulty nodes are detected and processes are migrated ahead of time.
Tuning of QoS Aware Load Balancing Algorithm (QoS–LB) for Highly Loaded Server Clusters
2001
This paper introduces a novel algorithm for content based switching. A content based scheduling algorithm (QoS Aware Load Balancing Algorithm, QoS-LB) which can be used at the front-end of the server cluster is presented. The front-end switch uses the content information of the requests and the load on the back servers to choose the server to handle each request. At the same time, different Quality of Service (QoS) classes of the customers can be considered as one parameter in the load balancing algorithm. This novel feature becomes more important when service providers begin to offer the same services for customers with different priorities.
On the Influence of PRNGs on Data Distribution
2012
The amount of digital information produced grows rapidly and constantly. Storage systems use clustered architectures designed to store and process this information efficiently. Their use introduces new challenges in storage systems development, like load-balancing and data distribution. A variety of randomized solutions handling data placement issues have been proposed and utilized. However, to the best of our knowledge, there has not yet been a structured analysis of the influence of pseudo random number generators (PRNGs) on the data distribution. In the first part of this paper we consider Consistent Hashing [1] as a combination of two consecutive phases: distribution of bins and distrib…
Cloud Infrastructure for Skin Cancer Scalable Detection System
2018
Skin cancer diagnostics is one of the medical areas where early diagnostic allows achieving patients’ high survival rate. Typically, skin cancer diagnostic is performed by dermatologist, since the amount of such specialists is limited, mortality rate is high [1]. By creating the low cost and easy to use diagnostic device, it is possible to bring skin cancer diagnostic to primary care physicians and allow to check much more persons and diagnose skin cancer on the early stages. There are several existing devices, that provide skin cancer diagnostics [2]. Most of them process the skin images locally and have limited diagnostic capabilities; some of them send images to dermatologists for manual…
An optimized path-selection using airtime metric in OLSR networks: Implementation and testing
2008
Wireless mesh networks used for broadband access may be equipped with multiple Internet gateways. Using airtime metric as the main criterion for path selection is an effective method for achieving load-balancing in such a multi-homed mesh network. In this paper, we describe how the radio-aware path selection using airtime as a metric can be implemented in order to achieve this functionality, based on a modified version of an open source OLSR implementation. The paper presents both the principle of this method, its implementation and the empirical results which focus on testing the airtime metric functionality based on a deployed small-scale test network.
The egoistic approach to parallel process migration into heterogeneous workstation network
1996
Abstract A new approach to the allocation of processes in a distributed system is discussed. The proposed solution deals with process migration into heterogeneous systems by means of a strategy that delegates the individual parallel applications to manage the migration of their processes by themselves, on the basis of their own performance objectives. This approach is discussed in opposition to the global scheduling based one, and the load balancing objective is pursued as an effect of the optimization of individual applications. A new performance evaluation criterion is introduced that consists in monitoring the delays that occur when two parallel processes run towards a common synchroniza…
A distributed-memory MPI parallelization scheme for multi-domain incompressible SPH
2022
A parallel scheme for a multi-domain truly incompressible smoothed particle hydrodynamics (SPH) approach is presented. The proposed method is developed for distributed-memory architectures through the Message Passing Interface (MPI) paradigm as communication between partitions. The proposal aims to overcome one of the main drawbacks of the SPH method, which is the high computational cost with respect to mesh-based methods, by coupling a multi-resolution approach with parallel computing techniques. The multi-domain approach aims to employ different resolutions by subdividing the computational domain into non-overlapping blocks separated by block interfaces. The particles belonging to differe…
3D high definition video coding on a GPU-based heterogeneous system
2013
H.264/MVC is a standard for supporting the sensation of 3D, based on coding from 2 (stereo) to N views. H.264/MVC adopts many coding options inherited from single view H.264/AVC, and thus its complexity is even higher, mainly because the number of processing views is higher. In this manuscript, we aim at an efficient parallelization of the most computationally intensive video encoding module for stereo sequences. In particular, inter prediction and its collaborative execution on a heterogeneous platform. The proposal is based on an efficient dynamic load balancing algorithm and on breaking encoding dependencies. Experimental results demonstrate the proposed algorithm's ability to reduce the…
Scavenger – A Framework for Efficient Evaluation of Dynamic and Modular Algorithms
2015
Machine Learning methods and algorithms are often highly modular in the sense that they rely on a large number of subalgorithms that are in principle interchangeable. For example, it is often possible to use various kinds of pre- and post-processing and various base classifiers or regressors as components of the same modular approach. We propose a framework, called Scavenger, that allows evaluating whole families of conceptually similar algorithms efficiently. The algorithms are represented as compositions, couplings and products of atomic subalgorithms. This allows partial results to be cached and shared between different instances of a modular algorithm, so that potentially expensive part…
Parallel Collision Queries on the GPU
2013
We present parallel algorithms to accelerate collision tests of rigid body objects for a high number of independent transformations as they occur in sampling-based motion planning and path validation problems. We compare various GPU approaches with a different level of parallelism against each other and against a parallel CPU implementation. Our algorithms require no sophisticated load balancing schemes. They make no assumption on the distribution of the input transformations and require no pre-processing. Yet, we can perform up to 1 million collision tests per second with our best GPU implementation in our benchmarks. This is about 2.5X faster than our reference multi-core CPU implementati…