LoadBuilder: a Tool for Generating and Modeling Workloads in Distributed Workstation Environments
Olivier Dalle
Abstract
Olivier Dalle
Abstract
This report presents LoadBuilder, a distributed environment designed to provide a portable way of empirically studying the effects of various kinds of workloads in local area networks of heterogeneous workstations. This tool is especially intended to build distributed experimentations including composite workload setting, statistics collecting and performance evaluations. The statistical analysis of the experimental results will help the dynamic load balancing designer to select the most meaningful indicators out of the plethora available on workstations, to establish behavior models of the workstations, to exhibit critical workload thresholds and finally, to establish his own set of meaningful workload indicators. In the following, we first describe the architecture of the environment. Then, we present and discuss the algorithms used to build (a) synthetic workloads, (b) statistics collectors and (c) measurement procedures.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This report presents LoadBuilder, a distributed environment designed to provide a portable way of empirically studying the effects of various kinds of workloads in local area networks of heterogeneous workstations. This tool is especially intended to build distributed experimentations including composite workload setting, statistics collecting and performance evaluations. The statistical analysis of the experimental results will help the dynamic load balancing designer to select the most meaningful indicators out of the plethora available on workstations, to establish behavior models of the workstations, to exhibit critical workload thresholds and finally, to establish his own set of meaningful workload indicators. In the following, we first describe the architecture of the environment. Then, we present and discuss the algorithms used to build (a) synthetic workloads, (b) statistics collectors and (c) measurement procedures.
Key concepts: Workstation, Workload, Computer science, Set (abstract data type), Architecture, Distributed computing, Operating system, Real-time computing