« Systematic Service Level Agreement (SLA) Data Management »

Abstract

In cloud computing, service resources are distributed. Cloud service markets respond simul- taneously to multiple, remote customers. Service exchange typically takes place over the Internet. Every customer needs to agree with a service level agreement (SLA) in order to lease a new service.

SLAs represent contractual terms and conditions between service providers and customers. Our work concentrates on the structured management of unstructured SLA data for cloud services. SLA information can be complex in terms of content and structure due to the diversity and plethora of offered cloud services.

Moreover, SLAs for cloud services lack standardization, which motivates their automated and consistent processing. We consider that SLA automation represents a prerequisite for any advancement around cloud service and SLA management. Accordingly, this work assumes that computing processes handle and exchange SLA contracts.

We propose the systematic management of SLA information through an SLA graph data model. The proposed SLA graph is derived from a solid specification for web service level agreements that was introduced by IBM research on utility computing [45]. The SLA graph is positioned in the cloud computing setting, where services are provisioned on-demand.

We analyze how flexibly the proposed SLA graph integrates data elements and relationships that are specified by diverse business domains. Moreover, we demonstrate the graph expres- siveness with respect to the representation and retrieval of complex service states and service element dependencies.

The SLA graph model supports the data exchange over connected environments and is not constrained by schema deployment or technical standard specifics. On the contrary, the proposed graph data model endorses any SLA unification and standardization effort for the cloud service domain.