Distributed Clouds: Modularity and Deployment Methods

点击翻译按钮获取中文摘要

Clients using cloud services are often caught unaware of some of their drawbacks.  These include: (1) the necessity to explicitly replicate services to accommodate zone failures, (2) the inability to have fine grain control over location of data, and (3) the fact that large megadata centers may reside in areas subject to vastly different privacy laws. Despite these drawbacks, clouds provide affordable, effective, dynamically scalable infrastructure vital to meet the needs of growing user populations and massive data sets. Almost every dimension of society relies on today's cloud services, including research, government, industry, healthcare, education, citizen science, and entertainment.******The proposed research program innovates with new methodologies for modularizing and deploying applications that will rely on a hybrid of cloud computing and distributed systems.  Key research initiatives will involve establishing accuracy when dynamically simulating system-level resources to reveal critical characteristics in deployment strategies, tradeoffs in the software engineering costs and benefits encountered, and practicality and sustainability in the context of real-world workloads.  Requirements will be derived from current needs that are of national and international significance. The proposed research program will include interdisciplinary teams from two different organizations at the University of Victoria, CanAssist and Oceans Networks Canada.  These organizations are currently identifying the needs of monitoring systems in the context of both healthcare and emergency response, and are committed to exploring solutions that will overcome the drawbacks associated with today's clouds.******By revealing more of the underlying platform and offering developers this "break in the clouds", cloud-based applications will have more control over not only the placement of data, but the quality of service provided by the supporting infrastructure. The key challenge will be to determine if these changes can be introduced to the existing cloud paradigm, without overwhelming application developers with complexity.**