Cloudscale's CloudStore

Times - 15:45 - 16:15 12 March Networking Coffee & 12:30 - 13:30 13 March Networking

The CloudScale CloudStore is:

  1. an example SaaS scenario for designing a scalable, elastic, and efficient online shop and
  2. an example migration scenario from a legacy book shop application to a scalable, elastic, and efficient cloud-aware online shop.

For these scenarios, the CloudScale project reused ideas from the TPC-W benchmark, a legacy benchmark of an online book store.

The benefits of CloudStore are:

  1. open and accessible

a.      source code (currently, Java Servlets and Java Spring) on Github. Development stack includes:

·         SpringFramework, Tomcat/Glassfish, SQL (Hibernate), NoSQL (MongoDB)
·         non-cloud legacy code uses JSF

b.     deployment scripts (currently, for EC2 and OpenStack)
c.      load generation scripts (currently, Open source web service for load generation supporting JMeter scripts)
d.    deployment scripts (currently, for distributed JMeter on AWS)
e.      analysis models (currently, Palladio models)

(2)    comparable results of

a.       analysis tools (e.g., CloudScale’s Analyser and Spotter)
b.      reverse engineering tools (e.g., CloudScale’s Extractor)
c.       best and worst practices (e.g., CloudScale's engineering method including best practice and anti-patterns)

(3)    a scenario for typical web applications (highly relevant class!)

The current limitations of CloudStore are:

  1. currently, focus on Java only, therefore, no support for non-java-based PaaS
  2. currently, only operates in IaaS/PaaS such as Amazon EC2 and OpenStack