6. BPMS Reference
Architecture
BPMS Architecture Survey
Thesis Structure
08 - November - 2016 6
RQ1
What is the best architecture of a
BPMS that facilitates a party switch in a
collaboration?
Runtime Party Switch
Technique
Collaboration Adaptation
Survey
RQ2
How can the BPMS of the parties
involved in a collaboration
support a runtime party change if the
collaboration
is controlled globally?
Runtime Party Switch
Technique without a
Global Controller
(Correlation Miner)
RQ3
How can a process discovery technique
facilitate a runtime party change in a
collaboration if the collaboration is
not controlled globally?
8. • BPMS-RA
• Available research studies in the BPM community, and
• Existing architectures in the literature and practice
• Result
08 - November - 2016 8
BPMS Reference Architecture
BPMS Architecture
Survey
Correlation Miner
BPMS Reference
Architecture
Collaboration
Adaptation Survey
Runtime Party Switch
Technique
9. • Systematic Literature Review (SLR) Methodology
• 78 selected articles out of 2,514 articles
• Results
• Runtime adaptation within a single-organizational has been well understood,
but…
• Further research must be undertaken to investigate the effect of runtime
adaptation on the inter-organizational collaboration setting.
08 - November - 2016 9
Runtime Party Switch Technique
BPMS Architecture
Survey
Correlation Miner
BPMS Reference
Architecture
Collaboration
Adaptation Survey
Runtime Party Switch
Technique
10. • A technique for enabling runtime party switch
1. replaces an old party in an inter-organizational collaboration with a new party,
2. determines the extent to which the state can be transferred from the old party
to the new party and transfers the state, and
3. communicates with other parties which may need to adapt their states as well.
• Evaluation
• Conceptual prototype, and
• Practical prototype in the context of the GET Service EU project
08 - November - 2016 10
Runtime Party Switch Technique
BPMS Architecture
Survey
Correlation Miner
BPMS Reference
Architecture
Collaboration
Adaptation Survey
Runtime Party Switch
Technique
11. • Two-Phase process discovery algorithm to
• Phase (1): reconstruct the missing collaboration model of the involved parties,
and
• Phase (2): relate the communications that belong to each other and, therefore,
belong to the same instance of a collaboration.
• Evaluation
• Synthetic event logs, and
• Real-world event log
08 - November - 2016 11
Correlation Miner
BPMS Architecture
Survey
Correlation Miner
BPMS Reference
Architecture
Collaboration
Adaptation Survey
Runtime Party Switch
Technique
12. Thank you for your Attention!
08 - November - 2016 12
BPMS Architecture Survey
BPMS Reference Architecture
Collaboration Adaptation Survey
Runtime Party Switch Technique
Correlation Miner
Future Work …
Editor's Notes
An organization performs a set of activities, often in a predefined structure, in order to realize a business goal.
When multiple organizations cooperate to achieve a business goal, this setting is called inter-organizational collaboration and each involved organization is a party in the collaboration.
The main issue here is, often, none of the involved parties have a full control on the collaboration since the parties, generally, do not want to expose their internal activities to each others.
However, during the execution a collaboration, which is being illustrated the yellow tokens, a party may drop out for technical reasons or for business reasons. In such a case, the leaving party must be replaced, at runtime, by a new party.
For example, in the transportation domain it is quite common that a freight forwarder switches from one carrier to another carrier. This switch can be a complex operation, in terms of the information that must be kept and transferred from the old carrier to its replacement. Such information may include administrative information regarding the cargo and the state of activities that must be performed on that cargo, such as whether the cargo has been picked up, or whether the cargo has been loaded into a container.
Given this example, not only the freight forwarder, but also other participating parties in a collaboration may be affected because of possible direct or indirect interaction between them and the leaving party. Therefore, the new party, ideally, must pick up where the leaving party has left off. But, what does that mean and why this is so challenging?
Because if we zoom into the activities that has been performed internally by the leaving party, we will see some differences between them and the internal activities of the new party. Therefore, probably, some tasks that have already been performed by the leaving party need to be cancelled, and some extra tasks need to be performed.
In this research, we assumed that each party employs a so-called Business Process Management Systsme (BPMS) to ensure that his activities are properly executed and monitored and, therefore, we formulated the main research questions as follows.
In order to answer this research question properly, we have decomposed it into three sub-questions.
The first research sub-question (RQ1) has focused on the best software architecture for Business Process Management Systems and it resulted into two artefacts, which are: a BPMS architecture survey and a BPMS reference architecture.
RQ2 has considered a situation in which it is possible to have a party as in a collaboration a global controller which observes all communications among all the involved parties and it resulted into two artefacts: a collaboration adaptation survey and a runtime party switch technique.
Finally, RQ3 considered a situation in which there is no such global controller in a collaboration. Therefore, the model that represents the collaboration does not exist. Hence, this research sub-question examined whether a process discovery technique can be used to capture the missing model of the collaboration, given that the participating parties have performed their work fully autonomously. This sub question resulted into one artefact which is a process discovery technique called the correlation miner.
We have surveyed the existing BPMS architectures in order to identify, classify and analyze the state-of-the-art in the components that constitute a BPMS in general and an adaptive BPMS in particular.
For this purpose, we have conducted a Systematic Literature Review. The review protocol in this study resulted in 41 selected articles out of 608 studies.
Finally, the results obtained by categorizing the selected articles into the BPMS architecture framework were presented. These results, for example, have revealed that:
– the minority of these architectures (i.e., around 30%) have covered the concept of inter-organizational collaboration, and
– the majority of architectures proposed by the selected articles (i.e., around 60%) were composed from scratch and therefore they are not based on existing architectures.
In order to cover the lack of reference architecture for BPM systems, subsequently, we have proposed a reference architecture for such systems, called BPMS-RA.
This referenc architecture has been designed based on the selected architectures as well as existing research in the BPM community. The BPMS-RA at the
coarse-grained level consists of three main components,
The SOA-WfMS component offers functionalities for designing and executing business processes
the BPI&BPA component is responsible for monitoring and decision making for a BPMS-RA-compliant system, and
the AAA component secures such systems by enabling authentication, authorization and accounting functionalities.
We have analyzed state-of-the-art techniques for dynamic adaptation in inter-organizational collaborations. We have again used a systematic literature review methodology in which 78 article were selected out of more than 2500 hits.
The findings from this analysis have highlighted the potential need for further research on the effect of changes at the inter-organizational level at the execution time.
Subsequently, we have developed a technique that enables runtime party switches in inter-organizational collaborations when they are controlled globally. This technique consists of three main steps:
Step1. replaces an old party in an inter-organizational collaboration with a new party,
Step2. determines the extent to which the state can be transferred from the old party to the new party and transfers the state, and
Step3. communicates with other parties which may need to adapt their states as well.
This technique was proposed under the assumption of having a global controller.
So, to release this assumption, we have developed a process discovery technique, to facilitate runtime party switches in inter-organizational collaborations when there is no global controller that observes communications which is called the correlation miner.
The correlation miner reconstructs the exchanged message and history of the performed activities of the collaboration via two phases.
Firstly, it captures the collaboration model of the involved parties based on their recorded communications and,
secondly, it relates the communications that belong to each other and, therefore, belong to the same instance of a collaboration.
Finally, we have performed many experiments on both synthetic and real-world execution logs. The results have shown the applicability of the correlation miner.