EMSE 2014 - The 1st International Workshop on Elasticity Mechanisms for Software-Defined Environments

Using software methods and tools to define, provision, control and optimize services and infrastructures, in large-scale, complex distributed systems based on dynamic multiple-dimensional requirements on the fly has been increasingly researched and applied. Such services and infrastructures are referred to as “Software-Defined Environments” (SDEs). Currently, this term typically refers to a collective approach in managing a computing infrastructure; IBM defines SDEs as:
“A Software Defined Environment (SDE) optimizes the entire computing infrastructure?compute, storage and network resources?so that it can adapt to the type of work required.”. We believe that SDEs are general concepts that cover integrated software methods and tools that produce intelligent management of multiple types of resources, be they machine-based (compute, networking or storage) or human-based (crowds or professional teams), for any software applications/services, platforms, middleware and infrastructures.
The notion of SDEs is thus closely related to multi-dimensional elasticity principles: the demands for dynamic provisioning and changes can be resource, quality and cost elasticity and their combination. Distributed Computing Infrastructures such as Cloud computing, ad-hoc wireless networks and P2P systems possess inherent abilities to support elasticity. Nevertheless, in most of today’s environments, resource elasticity is the main focus., Furthermore, its “throttling” is usually performed manually: Users are required to define the conditions for scaling up or down resources. Proper scaling conditions, especially when the application is executed on a third-party infrastructure are hard to define. Client needs change dynamically, requiring different optimizations relative to the amount of reserved resources. Finally, scaling relative to a single resource (e.g., compute, storage or networking elements) or a single dimension of elasticity (cost, quality and resource) leads to suboptimal use and possible performance degradation. To harvest the benefits of elastic provisioning and move towards the realization of SDEs, it is imperative that automated, multi-dimensional elasticity mechanisms be defined on multiple levels of a computing ecosystem. This central goal will be the focus of this workshop.