Support Load Balancing Research Articles

An Adaptive Resource Management Architecture for Active Networks

This paper presents a framework for resource management in highly dynamic active networks. The goal is to allocate and manage node resources in an efficient way while ensuring effective utilization of network and supporting load balancing. The framework supports co-existence of active and non-active nodes and proposes a novel Directory Service (DS) architecture that can be used to discover the suitable active nodes in the Internet and for selecting the best network path (end-to-end) and reserving the resources along the selected path. Intranode and internode resource management are facilitated through the DS, while within an active node the framework implements a composite scheduling scheme to schedule both the CPU and bandwidth resources to resolve the combined resource scheduling problems. In addition, a flexible active node database system and a simple adaptive prediction technique have been introduced in order to resolve the challenging problem of determining the CPU requirements of the incoming packets.

OVM: Out-of-order execution parallel virtual machine

High performance computing on parallel architectures currently uses different approaches depending on the hardware memory model of the architecture, the abstraction level of the programming environment and the nature of the application. In this article, we introduce an original client–server execution model based on RPCs called out-of-order parallel virtual machine (OVM). OVM aims to provide three main features: portability through a unique memory model, load-balancing using a plug-in support and high performance provided by several optimizations. The main optimizations are: non-blocking RPCs, data-flow management, persistent and non-persistent data, static data set distribution, dynamic scheduling and asynchronous global operations. We present OVM general architecture and demonstrate high performance for regular parallel applications, a parallel application with load balancing needs and a parallel application with real-time constraints. We firstly compare the performance of OVM and MPI for three kernels of the NAS 2.3. Then we illustrate the performance capability of OVM for a large real-life application that needs a load balancing support called AIRES. Finally, we present the performance of a real-time version of the PovRay ray-tracer demonstrating the reactiveness of OVM.

LinuxDirector: A connection director for scalable internet services

LinuxDirector is a connection director that supports load balancing among multiple Internet servers, which can be used to build scalable Internet services based on clusters of servers. LinuxDirector extends the TCP/IP stack of Linux kernel to support three IP load balancing techniques, VS/NAT, VS/TUN and VS/DR. Four scheduling algorithms have been implemented to assign connections to different servers. Scalability is achieved by transparently adding or removing a node in the cluster. High availability is provided by detecting node or daemon failures and reconfiguring the system appropriately. This paper describes the design and implementation of LinuxDirector and presents several of its features including scalability, high availability and connection affinity.

CRAUL: Compiler and Run‐Time Integration for Adaptation under Load

Clusters of workstations provide a cost‐effective, high performance parallel computing environment. These environments, however, are often shared by multiple users, or may consist of heterogeneous machines. As a result, parallel applications executing in these environments must operate despite unequal computational resources. For maximum performance, applications should automatically adapt execution to maximize use of the available resources. Ideally, this adaptation should be transparent to the application programmer. In this paper, we present CRAUL (Compiler and Run‐Time Integration for Adaptation Under Load), a system that dynamically balances computational load in a parallel application. Our target run‐time is software‐based distributed shared memory (SDSM). SDSM is a good target for parallelizing compilers since it reduces compile‐time complexity by providing data caching and other support for dynamic load balancing. CRAUL combines compile‐time support to identify data access patterns with a run‐time system that uses the access information to intelligently distribute the parallel workload in loop‐based programs. The distribution is chosen according to the relative power of the processors and so as to minimize SDSM overhead and maximize locality. We have evaluated the resulting load distribution in the presence of different types of load – computational, computational and memory intensive, and network load. CRAUL performs within 5–23% of ideal in the presence of load, and is able to improve on naive compiler‐based work distribution that does not take locality into account even in the absence of load.

GenoMap: A distributed system for unifying genotyping and genetic linkage analysis

This paper describes GenoMap, an integrated, parallel/distributed computing software system that aids in large-scale gene linkage mapping studies. This problem requires data from heterogeneous, distributed sources. In addition, the users of the system are researchers spanning a wide range of cultures, and represent a broad range of computing needs. A distributed, heterogeneous solution addresses most of the primary challenges, but presents its own difficulties. The key elements of GenoMap are provided by way of a web-based set of interactive Java applets and applications, as well as a small number of native applications, with automated load-balancing support for sharing computationally intensive tasks among a collection of distributed UNIX workstations. While computationally intensive tasks are supported on the most appropriate systems, users are able to conduct most of their work in familiar MacOS and Windows/PC environments through the use of a Java-enabled web browser. We also describe GenoScape, a C & X-Windows/UNIX native application for robust genotyping of a wide range of electrophoresis gel formats.

Run-time load balancing support for a parallel multiblock Euler/Navier-Stokes code with adaptive refinement on distributed memory computers

This paper describes the parallel implementation of algorithms requiring run-time load redistribution with the aid of the parallel programming library loco. As a typical application, a 2D finite volume multiblock Euler/Navier-Stokes code with block-wise adaptive mesh refinement is discussed. The loco software handles the communication between blocks and the distribution of blocks among the processors, thereby performing automatic load balancing at run-time. The loco library is interfaced with both the native NX communication primitives on Intel iPSC hypercubes and the PVM software on workstation clusters. The parallel performance of the code on the Intel iPSC/860 and on a DEC Alpha workstation cluster is discussed. In particular the effects of mesh refinement on the load balance are investigated.

A load balancing system for unix-based local area networks

This paper presents a distributed load balancing computer system developed for Unix-based local area networks. The system has been initially developed for supporting efficient distributed processing of knowledge-based multistage decision-making but it can also support load balancing in a general environment. The load balancing features were added to the Unix operating system without any modification of its kernel.

Voltage balancing using switched capacitors

Thyristor switched capacitors (TSCs) are a useful means of load balancing and reactive power support. The control of a general load by a binary weighted switched capacitor system is discussed in this paper along with a simple control scheme where the capacitor bank across a line pair is adjusted to control the fundamental voltage magnitude for that line pair. In a three-phase three-wire system, it is shown that an ideal reactive source can control positive sequence magnitude and cancel any negative phase sequence terms. With the quantized control levels available in a TSC, a ramping scheme with input hysteresis is shown to be effective in balancing in simulations and in a laboratory test with three banks in each phase. Low transient switching is employed with binary weighted capacitor banks to give a smooth variation of the reactive power supplied.