Skip to main content

Highly Available and Scalable Cloud Services

Our group develops novel techniques that enable cloud computing services to continue to provide good service in spite of inhabiting a challenging operating environment that presents unforeseen changes. Our techniques support cloud services to adapt to inevitable failures and changes in demand without requiring manual intervention.

Our goal is to improve the availability characteristics of applications and services deployed within the IaaS and PaaS offerings of IBM. Our efforts currently focus on three contributions. First, we work on support for dynamic binding of clients to cloud services to improve the resiliency of applications that have been developed as a composition of cloud services. Second, we investigate mechanisms for dynamic routing among cloud services in support of various use cases. Finally, we are developing a configurable health-management service that supports cloud services to maintain a desired QoS through fast and fine-grained failure detection and notification.