Version 32 (modified by bartek, 12 years ago) (diff)

--

QosCosGrid

Introduction

QosCosGrid could be viewed as a quasi-opportunistic supercomputer whose computational performance exceeds the power offered by a single supercomputing or data center. QosCosGrid is designed as a multi-layered architecture that is capable of dealing with computationally intensive large-scale, complex and parallel simulations that are usually too complex to run within a single computer cluster or machine. The QosCosGrid middleware enables computing resources (at the level of processor cores) from different administrative domains to be combined into a single powerful computing resource via Internet. Clearly, bandwidth and latency characteristics of the Internet may have an affect on overall application performance of QosCosGrid-enabled applications. However, the ability to connect and efficiently control advanced applications executed in parallel over the Internet is a feature that is highly appreciated by QosCosGrid users.

QCG Middleware

In a nutshell, the QosCosGrid middleware consists of two logical levels: grid domain and administrative domain. Grid-level services control, schedule and generally supervise the execution of end-users applications, which are spread between independent administrative domains. The administrative domain represents a single resource provider (e.g. HPC or datacenter) participating in a certain Grid or Cloud environment by sharing its computational resources, e.g. computing clusters, with both local and external end-users. The logical separation of administrative domains corresponds with the fact that they are possessed by different institutions or resource owners. Each institution contributes its resources for the benefit of the entire Grid or Cloud, while controlling its own administrative domain and own resource allocation/sharing policies.

The key component of every administrative domain in QosCosGrid is the QCG BES/AR service?, which gives the remote access to queuing systems resources. QCG BES/AR supports advance reservations, parallel execution environments - OpenMPI, ProActive and MUSCLE with coordinators responsible for synchronization of cross-cluster executions (see Cross-cluster communication) and QCG Data Movement services for managing input and output data. The another relevant service at the administrative domain is in charge of notification mechanism and it is called QCG Notification?. These services are tightly integrated and connected to the Grid-level services. The critical service on that level is QCG Broker?, which is a meta-scheduling framework controlling executions of applicaitons on the top of queuing systems via QCG BES/AR services. QosCosGrid architecture is depicted in the figure presented below.

No image "QCG-Architecture-v4.png" attached to main_old

End-user Information

The demanding nature of scientific simulations requires a new e-Infrastructure that is able to simultaneously manage heterogeneous resources, such as computing resources, storage and network to guarantee the level of Quality of Service (QoS) required by end-users for their applications, especially a large number of legacy applications designed to run in parallel. Addressing the need of end-users to run large scale complex simulations we have built a system capable to bring supercomputer-like performance for advanced applications, including sophisticated parameter sweep experiments, workflows, and more importantly cross-cluster parallel computations. If you are interested in using QCG as a user see this section.

Developer & Admin Information

Community

News

Download & Install

Interoperability & standards supported

Publications & Talks