Comparison of Performance Metrics of Media Applications

on Different Cloud based Host Servers

Group 11Arun Shivaramakrishna(#69127904)

Prateek Ramachandra(#28321312) Alok Boopalam Anantha Krishna(#91527264)

Motivation & Goal:

• Several media services like transcoding are nowadays being implemented over cloud platforms to achieve scalability.

• This project aims to compare the performance metrics of media applications deployed on two different cloud based platforms namely AWS and Google Compute Engine.

Related work

The following papers/articles compare different cloud platforms in general. Our work is focused on media applications alone.

• Comparing different cloud providers- https://www.cs.duke.edu/~angl/papers/imc10-cloudcmp.pdf

• Comparison of cloud platforms : Google cloud and AWS : http://cloudacademy.com/blog/google-cloud-vs-aws-a-comparison/

• Cloud computing platforms – metrics comparisons - http://www.researchgate.net/profile/Nawsher_Khan/publication/228067937_Cloud_Computing_Analysis_of_Various_Platforms/links/09e414feaf48f004db000000.pdf

• Performance analysis based on leading cloud computing services - http://www.cse.wustl.edu/~jain/cse567-11/ftp/clouds/

Implementation Details

• Using Amazon Elastic Transcoder, readily available with AWS, we uploaded a media file to access it over different devices like phones, tablets, etc across different formats .

• The problem we faced with using the Elastic Encoder was that the same service could not be used on another cloud platform(like Google Compute Engine) for comparison purposes, as its use is exclusive to the AWS platform.

• Thus, this inability to deploy the same media application over two different platforms using a common service led to us choose an OpenCV based media application to ensure the same media application code could be deployed on different platforms.

Testing and Evaluation Plan

• With a large data set as a benchmark, we plan to provide a comparison table which shows the conversion time from one format to another per 1 MB of data.

• We will explore different metrics for comparison.