Testing Cloud Failover
Roussi Roussev, VMware

The elastic properties of cloud computing introduce change. While extremely beneficial in terms of optimal resource utilization and agility, ch
ange can be dangerous. In this talk, I will describe several recent multi-day cascading failures and propose practical approaches to dealing with the problem. With the help of virtual machines, one can easily inject failures at device, host, cluster or datacenter level. And even if an issue slips through, multi-cloud environments can provide the necessary isolation and independence.

Roussi Roussev is a software engineer in the Datacenter Platform Group at VMware, where he works on configuration management and security solutions for VMware vSphere. Previously, Roussi held engineering position at Google where he worked on systems infrastructure, service monitoring and management. At Microsoft Research, he developed novel systems management, spyware and rootkit detection techniques, and helped build the most comprehensive client-side honeypot to identify malicious websites exploiting browser vulnerabilities. Roussi pursued graduate studies in computer sciences at Florida Institute of Technology, where he researched and developed model based testing, fault injection and malicious software detection methods. His interests include operating systems, security, testing and building large scale systems.