XLcloud: Design, Develop and Integrate The Software Elements of a High Performance Cloud Computing (HPCC)

Overview

XLcloud aims to define and demonstrate the principles of HPC as a Service (High Performance Computing) for all those applications that involve highly intensive calculations. XLcloud is designed as a collaborative tool that enables users to work together on highly sophisticated software in the Cloud, thus sidestepping the need for individuals to have to purchase expensive software on their own. XLcloud combines the expertise of companies and academics that are innovative in the field of high performance computer architectures and flow visualization HD/3D and video.

While many Internet-based applications are now available on the Cloud, high-performance applications still face a number of technological obstacles before they are cloud ready. For example, high performance applications often need large capacity processing, lots of storage and huge network bandwidth, needs that are difficult to meet in today’s Cloud.

In developing HPC as a Service (HaaS), XLcloud has identified the following high performance applications involving research and industrial innovation:

  • Research environments such as universities or technical institutes where high performance computing is a requirement,
  • Online gaming (serious games) and simulation,
  • Low-latency interactive video streaming,
  • Data-intensive HPC-based applications,
  • Online three-dimensional imaging for medical use.

Many of the above applications are inaccessible to small- and medium-size enterprises (SMEs) because of the levels of investment required and the budget pressure in IT. However, the model of cloud computing, by pooling resources and leveraging economies of scale, would allow these applications to find a true economic viability, create new ecosystems of suppliers and generate new uses to serving individuals and businesses. In the diagram below, we can see that XLcloud provides a collaborative platform that enables power users, business users as well as end users to work on and with highly sophisticated software running in a cloud.

Architecture-Diagram-8a.jpg

In brief, XLcloud believes that high-performance applications could benefit from the Cloud if certain technological obstacles were overcome: 

1. Functional diversity of High Performance Applications: Implementation of such apps requires very different architectures, ranging from real time to certain batch operations. Under these conditions, the deployment of the apps on a cloud infrastructure could only be economically viable as a service, if the shared infrastructure took into account the fundamental differences in architecture. As a project, XLcloud would establish a shared infrastructure like HaaS, in which the user could then instantiate Virtual Clusters. In so doing, it would provide the capacity for versatility allowing the Virtual Clusters to support the various models of architecture that are required by the users. 

2. Complexity: High Performance Applications are often required to handle complex data, especially in the field of imaging and video streaming. The operation of these applications in Cloud mode requires transferring the data stream composition onto the service-provider servers, and then returning these transactions to the user within their browsers. Constraints to this process are considerable, especially at the level of communication protocols and bandwidths. XLcloud aims to provide industrial solutions for these technical issues, which we believe are surmountable through the exploitation of new generation visualization applications in Cloud mode. 


In summary, XLcloud aims to remove some of the major obstacles to the operational implementation of high performance applications in the Cloud. Our goal is to enable members of our project ecosystem to take leadership positions in the market, as service or technology providers in the Cloud. In the diagrams below, we present, as an example, some of the workflows involved in deploying and maintaining an instance of a Remote Rendering Virtual Cluster and how a user can deploy a 3D interactive application on top of it.

Remote Rendering Virtual Cluster Deployment and Auto-Scaling

RRVC Arch Diagram2.png 

A user deploys an application in the Remote Rendering Virtual Cluster

RRVC Arch Diagram3.png

Handling of Remote Rendering Sessions

RRVC Arch Diagram1.png


This wiki is licensed under a Creative Commons 2.0 license
XWiki Enterprise 4.5.1 - Documentation - Legal Notice

Site maintained by