Article Preview
TopIntroduction
An organization, whatever its size, typically has to manage a set of business processes (BPs) in different sectors such as human resources, logistics, marketing, etc. A BP is a collection of interrelated tasks that should be executed to accomplish well-defined goals (Weske, 2019) (WfMC, 2019). Tasks are the most basic (atomic) element and are part of activities, which in turn can be part of sub-processes.
Outsourcing a process consists in entrusting it totally or partially to a third party who takes charge of its execution (Rekik et al., 2016). This practice has appeared in the 90’s when companies exchanged their processes with each other according to their competencies (Rekik et al., 2015). Nevertheless, in recent years, the most chosen outsourcing environment has become the Cloud computing, which according to the National Institute of Standards and Technology (NIST) (Mell & Grance, 2011) is a model for enabling ubiquitous, convenient, and on-demand network access to configurable computing resources (e.g., networks, servers, storage, applications and services).
The Cloud computing lures companies with its new economic model of pay-per-use and its elasticity that allows to adjust the resource consumption according to the demand. However, the Cloud is often criticized for security issues and loss of control over data and processes. Therefore, when outsourcing processes to the Cloud, it is highly recommended for a company to distribute each of its processes over several Cloud providers in order to protect its know-how and take advantage from the strengths of each offer (Goettelmann et al., 2013) (Ahmed Nacer et al., 2018). Indeed, as shown in Figure 1, the selected providers will have only a partial view of the process and will not be able to understand how it really works.
However, if companies exaggerate in the distribution, each task of an outsourced process will be assigned to a separate provider. So for example a simple process of six tasks will end up with six different Cloud providers. This may significantly increase the overall cost and time of the process due to data transfers between tasks. This is why a company will have to find the right balance for its processes according to their requirements, knowing that the cost and performance vary according to the tariffs, bandwidths and calculation capacity of Cloud providers.
Figure 1.
Distribution of a process over several Cloud providers
In this context, several work like (Goettelmann et al., 2014; Bhushan & Reddy, 2018; Guidara et al., 2016) focus on business process outsourcing (BPO) and service selection but most of them are not comprehensive enough in terms of criteria and neglect some very important ones such as data transfer and process obfuscation. In addition, some characteristics of Cloud computing like availability are not always considered.
To overcome these limitations, we propose in this paper a selection method based on a heuristic algorithm that helps companies in combining the best Cloud offers for their processes. The proposed algorithm allows to find the right balance between cost and performance while guaranteeing process obfuscation based on a set of criteria, namely cost execution, runtime, data transfer, task separation and delay margin. Moreover, the algorithm takes into account the company's requirements by weighting the cost and performance criteria according to the tolerated delay.
The proposed selection method is mainly intended for the first migrations to the Cloud and complements our decision-making support (Zarour & Benmerzoug, 2019) which is much more oriented towards security and compliance as well as our Business Intelligence approach (Zarour & Benmerzoug, 2017) that aims to analyze the execution history accumulated on outsourced processes.