A Cross-Layer BPaaS Adaptation Framework Kyriakos Kritikos, Chrysostomos Zeginis Information Systems Laboratory ICS-FORTH, Heraklion, Crete, Greece Email: kritikos@ics.forth.gr Frank Griesinger, Daniel Seybold, J¨ org Domaschka Institute of Information Resource Management Ulm University, Ulm, Germany Email: {fname.lname}@uni-ulm.de Abstract—The notion of a BPaaS is currently taking a mo- mentum as many organisations attempt to move and offer their business processes (BPs) in the cloud. Such BPs need to be adaptively provisioned so as to sustain the service level promised in the respective SLA. However, current cloud-based adaptation frameworks cannot cover all possible abstraction levels and usually rely on simplistic adaptation rules. As such, this paper proposes a novel BPaaS adaptation framework able to orchestrate actions on different abstraction levels so as to better address the current problematic situation. This framework can support the dynamic generation of adaptation workflows as well as the recording of the adaptation history for analysis purposes. It is also coupled with the CAMEL language which has been extended to support the specification of cross-level adaptation workflows. I. I NTRODUCTION In order to survive in such a dynamic business world, organisations need to exploit service-based business processes (BPs) which can adapt to the current situation as well as offer suitable service levels to their customers. As such, the cloud can be considered as one of the most suitable medium to support such capabilities as apart from promising the supply of infinite and cheap commodity resources, it also enables organisations to focus mainly on their core business (as, e.g., technical administration can be outsourced). As such, we are currently observing a trend of BP migration in the cloud which leads to the notion of Business Process as a Service (BPaaS). BPaaS indicates the delivery of BPs as on-demand services. Its paradigm enables well the role of a BPaaS broker, an entity able to either provide BPaaS products to potential organisations to realise their core or support BPs or supply consulting services to facilitate these organisations to move their business to the cloud. Such a broker, while might have great expertise in BP and workflow modelling, should be supported with tools or technical platforms able to cover the whole lifecycle [1] of the BPaaSes offered without requiring from the broker to dig into quite specific cloud technicalities. The BPaaS life cycle ranges from the BPaaS design by business experts down to its technical operation on cloud resources. In the European project CloudSocket 1 , the phases of this life cycle were defined as: (i) Design BPs at the highest level, (ii) Allocation as mapping from BPs to cloud services, (iii) Execution of the BPs based on the allocated cloud services, and (iv) Evaluation of the BPaaS instance (see Figure 1). To cater for this, CloudSocket developed an 1 www.cloudsocket.eu architecture built on this paradigm with one environment per life-cycle phase. The presented adaptation framework in this paper was integrated into the Execution Environment, as this is the phase were adaptation happens at run-time. Design Environment Allocation Environment Execution Environment Evaluation Environment Design Allocation Execution Evaluation BPaaS Fig. 1. BPaaS life cycle and the environments of the CloudSocket project. By focusing on the execution lifecycle phase, which in- volves executing as well as dynamically provisioning a BPaaS, there is a great need to support the adaptation of a BPaaS to sustain the service level promised. Such an adaptation cannot be simply in the form of simple event to scaling actions, a current capability of existing cloud orchestration tools [2]. This form might be suitable in simple cases but is not sufficient to cover more sophisticated scenarios for BPaaS adaptation, subject to the fact that adaptation might require to occur in different abstraction levels, apart from the lower, infrastructure one, including the platform, service and workflow levels. By considering the service-oriented computing literature, we can fortunately see some novel cross-level frameworks able to support more composite adaptation scenarios. However, such frameworks usually cover at most two levels and were developed before cloud computing came into real play. In addition, they usually employ level-specific adaptation mech- anisms, triggered in an individual manner. In this way, the same situation could be sensed by these mechanisms which might attempt to react by performing respective adaptation actions. Without a central orhestration of such actions, there is a high risk that actions on one level might undo or interfere with actions on another level. This results in over-spending of resources as well as to vicious adaptation cycles. To solve the above problem, this paper proposes a novel BPaaS adaptation framework able to orchestrate level-specific adaptation actions across levels and clouds. This framework builds upon a novel extension of the CAMEL 2 cloud appli- cation language [3], [4] able to support the specification of 2 www.camel-dsl.org