Scaling Virtualized Smartphone Images in the Cloud
Name
Kristiina Ritso
Abstract
In this thesis we deployed a smartphone image in an Amazon EC2 instance and ran stress tests on them to know how much users can one instance bear and how scalable it is. We tested how much time would a method run in a physical Android device and in a cloud instance. We deployed CyanogenMod and Dalvik for a single instance. We used Tsung for stress testing. For those tests we also made a Tomcat server on Dalvik instance that would take the incoming file, the file would be compiled with java and its class file would be wrapped into dex, a Dalvik executable file, that is later executed with Dalvik. Three instances made a Tsung cluster that sent load to a Dalvik Virtual Machine instance. For scaling we used Amazon Auto Scaling tool and Elastic Load Balancer that divided incoming load between the instances.
Graduation Thesis language
English
Graduation Thesis type
Bachelor - Information Technology
Supervisor(s)
Satish Narayana Srirama, Huber Flores
Defence year
2013