Sunflower is an adaptive P2P agent-based framework for configuring, enacting, managing and adapting autonomic workflows. Sunflower assumes that multiple copies of a Web service, with different performance profiles and distributed in different locations, co-exist. During the execution of the workflow, if a service fails or becomes overloaded, a selfreconfiguring mechanism based on a binding adaptation model is used to ensure that the running workflow is not interrupted but its structure is adapted in response to both internal or external changes.

Workflows are described in Sunflower by the BPEL language in order to exploit existing design tools. Sunflower replaces the standard BPEL engine with a new decentralized engine able to exploit the dynamic information available in the Grid and respond to the dynamic nature of the Grid. The workflow process is enacted by a set of cooperating Sunflower BPEL engines (SBE), instantiated at all participating nodes, who are responsible for interpreting and activating part of the process definition and interacting with the external resources -invoked web services- necessary to process the various activities.

The Sunﬂower Console provides the user’s interface to model a workﬂow using the Eclipse BPEL Designer and gener-ate the XML-BPEL code. Through the decomposer module the XML-BPEL code is partitioned into sub-workﬂow schema that are associated to the SWEA agents. The Sunﬂower Console also enables the deployment of the workﬂow according the initial conﬁguration on the nodes of the Grid. During this phase the SWEA agents are queued on SBE nodes that contain the Web services to invoke. In order to mask the complexity of the underlying infrastructure two different environments must be installed. A standard Sunﬂower environment is installed on the Grid nodes. On the Cloud, Sunﬂower uses a Software as a Service (SaaS) delivery model which provides different customers the functionality of an application that is completely hosted in the cloud.

References

G. Folino, A. Forestiero, G. Papuzzo, G. Spezzano.
MOSE’: A grid portal for solving geoscience problems using distributed knowledge discovery services.
Future Generation Computer Systems, Elsever, Vol. 26, N. 1, pp. 87 – 96 (2010).

D. Cistaro, G. Papuzzo.
Manuale Sunflower - Un sistema autonomico per la gestione di workflow basati sul modello a coreografia per Grid e Cloud.
Rapporto Tecnico Interno “RT-ICAR-CS-10-11”, Istituto di Calcolo e Reti ad Alte Prestazioni (ICAR-CNR) del Consiglio Nazionale delle Ricerche.