Software defined environments for managing infrastructure services in cloud computing, comprising compute, storage, and network functionality, are an area of active research. This paper demonstrates the need to extend this research to managed services that are offered by IT (information technology) service providers who manage customers' servers and applications to meet functional (e.g., monitoring, patching, and backup) and nonfunctional (e.g., availability, resilience, and portability) requirements. To enable software defined control, managed services must be modularized, and their functionality must be exposed through application programming interfaces (APIs) for lifecycle management, composition, and configuration. This paper provides a vision for service modularization based on real-world customer requirements and describes the pros and cons of providing different classes of APIs from the perspectives of both customers and cloud providers. An example of a complex managed service using monitoring, event management, and incident management components, based on the authors' experiences with a real-world cloud implementation, is used to illustrate the key challenges involved in building software defined managed services. This is an emerging area of research that will enable continuous enhancement of IBM's cloud offerings to satisfy changing customer requirements.
[1]
David E. Culler,et al.
The ganglia distributed monitoring system: design, implementation, and experience
,
2004,
Parallel Comput..
[2]
Kornel Terplan,et al.
HP Openview: A Manager's Guide
,
1997
.
[3]
Valentina Salapura,et al.
Resilient cloud computing
,
2013,
IBM J. Res. Dev..
[4]
Yang Song,et al.
Efficient and agile storage management in software defined environments
,
2014,
IBM J. Res. Dev..
[5]
Nancy Alonistioti,et al.
Software Defined Radio: Architectures, Systems and Functions
,
2003
.
[6]
Wanjiun Liao,et al.
Software defined networks [Guest Editorial]
,
2013,
IEEE Communications Magazine.
[7]
Michael H. Kalantar,et al.
Weaver: Language and runtime for software defined environments
,
2014,
IBM J. Res. Dev..