BPaaS Evaluation Environment
Introduction
The BPaaS Evaluation Environment aims at performing an analysis over the information produced during the execution phase of the BPaaS lifecycle in order to discover bottlenecks as well as optimisation points. Such information can then be exploited in a next cycle by the BPaaS Design and Allocation Environments in order to optimize the BPaaS such that it is able to exhibit a particular service level as well as improve the business benefits of the CloudSocket Broker offering it.
To support such analysis, the BPaaS Evaluation Environment follows a multi-layer architecture comprising three main layers: (a) the UI layer which visualizes the results of the analysis; (b) the business logic layer offering the analysis functionality to the layer above, the UI one; (c) the data layer which provides all information required to support the analysis.
The main functional capabilities offered by the BPaaS Evaluation Environment are the following:
- aggregation and decomposition from technical logs up to domain-specific business key performance indicators (KPIs) - conceptual analytics enabling root cause analysis, drill-down of business KPIs and propagation of technical indicators - discovery of best deployments for BPaaSs - detection of event patterns leading to KPI or SLO violations and their mapping to adaptation rules to support both re-active and proactive BPaaS adaptation - semantic process mining to detect bottlenecks and places for improvement for both BPaaS business processes and workflows.
The core functional capabilities are supported via the following visualisation & data building functionalities:
- semantic lifting of process, service and monitoring logs - visualisation of a monitoring dashboard that presents a management overview for the BPaaSs running in the cloud
Layers
UI Layer
Hybrid Business Dashboard
The Hybrid Business Dashboard enables selecting the most suitable analysis techniques and visualises the respective analysis results. It also offers drill-down of business to technical KPIs facilitating the performance of a root cause analysis. In addition, it enables the broker to finalise incomplete KPI specifications.
The dashboard comprises three main views:
(a) the Business Indicator view enables the visualisation of business and technical goals & KPIs which are represented in different categories. Common representations like traffic lights will be used to indicate whether KPIs are met or violated. (b) The Process Deployment/Adaptation Analysis view will visualise in a table-based manner improvement potentials by presenting the findings of the respective analysis performed. There will also be capabilities to sort, search or the export in an appropriate format such findings. (c) The Process Mining view visualises the original process model by also assimilating the log mining results into this model in order to highlight deviations from the original design, bottlenecks as well as places for improvement.
Depending on the type of analysis required, the Hybrid Business Dashboard invokes the corresponding component at the next layer, the business logic one, that is the Conceptual Analytics Engine for KPI & best deployment / event pattern analysis and the Process Mining Engine for log mining analysis.
Business Logic Layer
Conceptual Analytics Engine
The Conceptual Analytics Engine is responsible for performing KPI analysis by issuing corresponding semantic queries at the underlying Semantic KnowledgeBase (SKB) of the next layer, the data one. It is also responsible for best BPaaS deployment discovery through the evaluation of semantic rules over the content of SKB. Such analysis can assist during the BPaaS allocation phase both for initially-designed BPaaSs or existing BPaaSs with performance issues. Another major functionality exposed is the detection of event patterns through the application of a logic-based mining approach over the SKB content. Such patterns can then be semi-automatically transformed into adaptation rules to be enforced by the Execution Environment in order to support the pro-active and re-active adaptation of BPaaS in the context of KPI or Service Level Objective (SLO) violations.
Process Mining Engine
The Process Mining Engine is responsible for mining process, workflow, service, infrastructure and monitoring logs in order to discover discrepancies between what has beend designed and what is being executed, specific process bottlenecks as well as places for improvement. It will not reinvent the wheel but exploit existing state-of-the-art mining algorithms which will be adapted to operate over an SKB.
Data Layer
Semantic KnowledgeBase
The Semantic KnowledgeBase (SKB) will draw information from the BPaaS Execution Environment in the form of process, workflow, service, infrastructure and monitoring logs which will then be semantically lifted, linked and stored. SKB will also comprise semantic rules operating over its semantic content which will enable the derivation of added-value facts which will support the various types of analysis that are required by the layers above.
MetaModel Platform
The MetaModel Platform enables the storage and retrieval of various models that are required for those types of analysis offered by the Evaluation Environment, such as process, workflow and KPI models. It also supports the functionalities of model export, transformation and graphical representation.
Funding Partners
Horizon 2020