Cloud Economics

Demystifying Cloud Computing Costs

With the proliferation of Cloud platforms, Enterprises are gearing up to develop and deploy Cloud based applications. One of the first questions they have to answer is “Does it make economic sense?” but when it comes to determining the Economic Impact of the cloud, it is an inexact science.

Complex granular numbers like $0.10 per compute hour / $0.01 per 10K messages makes it very hard to put it in perspective of real world applications and determine if the potential benefits are worth it and what the Total Cost of Ownership (TCO) is likely to be.

Cumulux is authoring a three part white paper series aimed at demystifying the pricing model of the Cloud in the context of typical application scenarios. It is positioned at decision makers, enabling them to make educated decisions about the Total Cost of Ownership of running applications on the Azure Services Platform.

Part I – Compute Intensive Application

In this first part, we look at a compute heavy Financial Services application running on Windows Azure.

Background
Customer Segment Financial Research
Application Commodities Market Analysis
Applicable Verticals Financial Services, Insurance,
Engineering, Healthcare, Bio-Tech

Application and Challenges

This application processes daily commodity pricing data and runs parallel simulations to predict the moving price averages. This compute intensive solution is severely limited by the linear scalability model of the existing data centers as customers have to plan and build for peak capacities regardless of actual processing throughputs. The application will run several long running parallel computations each of which runs simulations based on a large input data set.

Application Scenario

The application runs a number of simulations in parallel based on complex mathematical algorithms. On every market day (22 days/month) 100-300 simulations are done. Each simulation takes about 10-15 hrs hours to complete based on a single CPU 2 GB machine.

  • Input to the system is a 10 GB binary file reflecting the commodity markets data on a daily basis
  • Output from each of the simulations is a 200 MB data file
  • Storage needs are transient; Input data is provided everyday and the output of the analysis is dynamically downloaded to an on-premise SQL Server database
Simulations 50
Instance Charges $1,406.40
Bandwidth Charges
Data Transfer In $22.00
Data Transfer Out $33.00
Storage
Raw Storage $66.00
Storage Operations Cost $110.00
Total $1,637.40
downloads

Engage us to build a detailed Analysis for your portfolio