Self-managing distributed systems and globally interoperable network of clouds

While self-managing IT has been the target of every enterprise since the introduction of autonomic computing, it has eluded us till recently. While there has been much progress in using autonomic computing to model, configure, monitor and control the external world outside of the computing processes, today, IT management consumes 70% of the IT budget to keep the applications available, secure and compliant with regulations. For every dollar spent on software development, another $1.38 is spent on managing and maintaining it. Every time a fault occurs, one has to stop, isolate it, diagnose it and fix it and it requires an army of experts from different disciplines to do it. This becomes an impossible task when large scale, and fluctuations both in workload demands and available resource pools are involved. In this paper we describe a new approach that brings finally self-managing properties to applications and workflows using a new computing model. Using this approach, we have reduced IT complexity and created an interoperable global network of clouds that can be used to support self-provisioning workloads with auto-scaling, auto-failover and live migration without disrupting user experience.