analyse the cost implications of adopting a cloud-based HPC architecture using Amazon Web Services: Cloud computing Coursework, UOEL, UK

(a) Description

A given company is exploring a move to a cloud-based architecture for one of their key High Performance Computing (HPC) systems

1. They have asked you to help them analyse the cost implications of adopting
a cloud-based HPC architecture using Amazon Web Services (AWS) offerings. The company management

wants to know what the economic benefits of utility computing for HPC are, in comparison to their current on-premises setup.

Prepare a case study to quantify the cost benefits by comparing the economic benefits of the proposed cloud computing architecture vs. the traditional (on-premises) one.

The case study ought to be based on a real, medium or large company of a given industrial sector in Europe focused on a specific HPC application domain. For the purposes of this project, the following seven HPC application domains have been selected:

The actual HPC application domain you will focus on will be defined by a simple (hash) function based on your student id. To calculate it, add all digits of your student number and then apply the modulo operator on the seven sectors (%7) e.g. Mary Murphy with student id 20123457 will work on the “New materials modelling

(b) Assumptions

You are asked to design a multi-variable cloud/utility HPC architecture that realistically fulfils the different system requirements such as:

Functional: Computation, memory, and data storage needs, number of users, requests Non-Functional Expected SLA, QoS, and service/maintenance costs, and any other relevant details.

By using a utility model, you will simulate economic behaviours and compare traditional (on-premises) vs. utility computing (cloud-based) HPC architectures.

Use available AWS service offerings and prices for European AWS Regions only, including GPU and other offloading devices. Make sure that hidden costs are accounted for in your case study and analyse a 5-year horizon with all costs and any possible expansions.

• Describe any and all possible acceptance criteria other than direct costs: specific benchmark results (e.g. Linpack [ 2], TPC,…), response times, etc. You must conduct some independent research and include any relevant bibliography.

• State in your conclusions what the best option is. Illustrate an understanding of your clients business needs and how the adoption of a HPC solution would benefit the company