Cloud Computing Development

[toc]

Introduction

Cluster based systems have transformed much in past years. Starting with early GRID approach, the industry is experiencing another transformation toward cloud services since 2007.

It first started with simple, core, services by vendors offering infrastructure services (servers, storage etc.) and evolved into more complex solutions available today. One can find email, file systems, data bases and even application level solutions (built on cloud infrastructures) powered by cloud systems.

As internet bandwidth increased, the number of services can be offered has grown as well, opening the way for new innovative solutions, video based and rich multimedia.

Among the largest core cloud providers are: Amazon™ (AWS™) and Microsoft® Azure® through their cost-effective and rich solutions portfolio.


Why Go Cloud?

Public cloud environments offer many benefits over traditional way of thinking.

1. Hardwareless Operation

Cloud providers offer just the core systems and organization needs to operate: Servers and Storage. There is no need for internal IT maintenance or purchase of real hardware systems, but configure the cloud to work as you expect. Therefore initial hardware costs are eliminated.

2. Pay-per-Use / Pay-as-you-Grow

Usually, several operational plans are offered, so users can choose the one which best suits their expected usage. It is possible to adapt pre-allocation of resources and also fixed payments, but cloud models offer great economic flexibility.

3. Scalability

One can grow based on request, without the need to operate a fixed amount of computing systems, which is limiting in two main forms: paying more than necessary when services are not consumed and unable to grow once resources are at full capacity.
Advanced cloud control allows to dynamically add more systems or shutdown to maintain costs where expected.


Our Approach

Through our development of HPC systems, cloud solutions were taken to their peak offering.

With the introduction of GPU hardware (NVIDIA® Tesla® and more), another performance advantage was added for cloud systems.
Organizations in early development stages can enjoy low cost cloud plan to gain maturity and enhance skills, instead of purchasing expensive industrial GPGPU systems.

For example, Hoopoe™ was designed as a cloud platform and additional services, powering private (internal) clouds and also able to integrate with or run on top of Amazon™ AWS™ services.

We focus mostly around Amazon™ AWS™ and Microsoft® Azure® services to provide general computing (EC2), storage (S3 etc.), DB and HPC/GPGPU computing solutions over the cloud.

Development supports the following technologies:

Operating Systems

  • Microsoft® Windows® (Desktop / Server)
  • Linux (All supported distributions)

HPC Software Infrastructures

  • NVIDIA® CUDA®
  • OpenCL™
  • Graphics and Visualization: OpenGL / DirectX
  • Multimedia: Image and Video Processing (Decode/Encode etc.)
  • OpenMP (Multi-Core)
  • MPI (Message Passing Interface)
  • Accelerated Libraries and Performance Primitives (OpenCV, Intel IPP etc.)

Supported Hypervisors & Toolstacks

We offer our expertise in development and design of distributed, high performance, systems, targeting cloud or cluster based environments.

It is important to note that there are times where cloud based solution will not deliver expected performance or remain cost-effective.
With appropriate consulting and analysis, it is possible to avoid these cases.


More Information

Contact us: support@cass-hpc.com.