Continuous Integration, Deployment and Validation: Supporting Scalable Industrial Ecosystems

Continuous Integration, Deployment and Validation: Supporting Scalable Industrial Ecosystems Miguel A. Mateo-Casali* Research Center on Production Management and Engineering Universitat Politècnica de València Valencia, Spain [email protected] 0000-0001-5086-9378 Andrés Boza Research Center on Production Management and Engineering Universitat Politècnica de València Valencia, Spain [email protected] 0000-0002-5429-0416 Ciprian Candea Ropardo SRL Sibiu, Romania [email protected] 0000-0002-8869-0194 Francisco Fraile Research Center on Production Management and Engineering Universitat Politècnica de València Valencia, Spain [email protected] 0000-0003-0852-8953 Lukas Morand Fraunhofer Institute for Mechanics of Materials IWM Baden-Württemberg, Deutschland [email protected] 0000-0008-8566-7642 Yoav Nahshon Fraunhofer Institute for Mechanics of Materials IWM Baden-Württemberg, Deutschland [email protected] 0000-0003-4774-8764 Abstract— The DiMAT project (Digital Technologies for Modelling, Simulation and Optimisation in Materials) aims to deliver open-source digital tools to accelerate the digital transformation of small and medium-sized enterprises (SMEs). This article presents the design and implementation of Continuous Integration (CI) and Continuous Deployment (CD) pipelines, which support the development, testing, and delivery of DiMAT platform components. By leveraging Docker containerisation and Kubernetes orchestration, the platform ensures robust scalability, efficient resource use, and cloudnative modularity. Security is enforced through Keycloak-based identity management, while interoperability is supported via standards such as IIRA, RAMI 4.0, IDSA-RAM, EMMO, MODA, and CHADA. These integrated approaches enable secure and reliable data exchange across toolkits and ensure the continuous delivery of new features in industrial research environments. The proposed CI/CD architecture serves as a replicable model for other industrial digitalisation projects. Keywords: CI/CD, Continuous Integration, Continuous Deployment, Microservices, Containerisation, Interoperability, Industry 4.0, Cloud Architecture I. INTRODUCTION Digital transformation in the manufacturing industry, i.e. within Industry 4.0, has generated a paradigm shift in production processes. In this new scenario, incorporating advanced digital technologies becomes essential to optimise materials manufacturing processes, improve product quality and enable companies to respond rapidly to market demands [1]. The integration of advanced communication and deployment paradigms is fundamental to addressing the challenges of modern industrial systems. The current digital transformation era demands the adoption of robust architectures that can scale, adapt and ensure secure operations. Within this context, the development of the DiMAT architecture [2] serves as an example of an innovative approach to building a modular and distributed ecosystem that integrates toolsets and supports seamless communication between services and within them. The DiMAT architecture has been developed in accordance with the ISO/IEC/IEEE 42010 standard [3], which provides a comprehensive framework for the description and design of architectures. By adhering to the principles of this standard, the architecture ensures consistency in design practices and fosters interoperability among the various components of the system. Our aim is to provide a modular deployment design that supports current requirements and lays the foundations for future implementations. The present contribution introduces and showcases our deployment design strategy. The architecture is designed to be scalable and adaptable, ensuring that new functionalities and toolsets can be integrated with minimal disruption. This is especially important in environments where high reliability and robust security measures are critical for real-time data exchange. The DiMAT platform leverages modern RESTful API methodologies and traditional message queue paradigms to create a resilient and flexible communication infrastructure. Furthermore, integrating containerisation technologies, such as Docker, and orchestration tools, like Kubernetes, underscores our commitment to achieving high availability and scalability in a cloud-based deployment [4], [5]. Moreover, it is important to highlight that the architecture significantly emphasises security. As industrial environments become increasingly interconnected, the risk profile associated with cyber threats rises. To mitigate these risks, the architecture incorporates secure access protocols, API authorisation mechanisms and identity and access management solutions. For example, using Keycloak, a widely adopted open-source identity management system, ensures that the authentication and authorisation processes are robust and scalable. This integrated security framework is complemented by systematic backup and recovery strategies designed to protect against data loss and system failures [6]. In addition to these core elements, the DiMAT architecture places a strong emphasis on interoperability. The diverse tools and services integrated into the platform operate with shared terminology, standardised data structures and stable service interfaces. This facilitates monitored communication and ensures that the system can evolve as new technologies and requirements emerge. In essence, the architecture serves as a model for integrating secure, scalable and interoperable solutions into a cohesive ecosystem that can be utilised as a reference model for future deployments. The significance of the DiMAT project is not limited to its technical innovations. It also provides a methodological framework that can be replicated in other fields requiring a similar integration of digital tools and cloud-native services. Within this context, Continuous Integration and Continuous Delivery (CI/CD) methodologies have become established as fundamental tools for the agile and secure deployment of technological improvements. These practices facilitate constantly updating systems and processes, significantly reducing downtime and minimising the risks associated with introducing changes in production environments. The structure of this article is as follows: Section (i) introduces the context of digital transformation in manufacturing, illustrating how advanced digital technologies support scalability and secure operations. Section (ii) relates the DiMAT platform to existing theories and frameworks, discussing how standards like IIRA, RAMI 4.0, and IDSARAM inform its development. Section (iii) describes the research approach underpinning the platform’s modular and distributed design, highlighting the role of containerisation and identity management. Section (iv) presents the main findings of our work, focusing on the successful integration of Docker-based deployments, security measures, and CI/CD pipelines. Section (v) synthesises key conclusions and potential future directions to reinforce the DiMAT architecture’s applicability to industrial contexts. II. RELATION TO EXISTING THEORIES AND WORK The development of the DiMAT platform architecture responds to the need to provide an integrated digital ecosystem for the modelling, simulation and optimisation of advanced materials. This ecosystem has been designed to improve the efficiency of design and manufacturing processes while ensuring high levels of interoperability, security, scalability and ease of integration with external tools. Throughout its development, emphasis has been placed on the architecture's ability to evolve and adapt to new technologies and requirements without compromising its stability and functionality. Since its inception, DiMAT's architecture has followed a modular and distributed approach, allowing its components to operate autonomously while collaborating within a unified environment. Implementing international standards, such as ISO/IEC/IEEE 42010, has been key in structuring the design, ensuring that each component meets well-defined technical requirements and that all perspectives necessary for its correct implementation are addressed. This architectural design is based on fundamental principles that govern its structure and operation. These include horizontal and vertical scalability, containerisation and orchestration technologies such as Docker and Kubernetes, adopting standardised APIs for communication between components, and implementing a robust security and identity management system based on Keycloak. In addition, the principles of interoperability and standardisation have been followed by integrating modelling and material characterisation standards such as EMMO, MODA and CHADA. This section describes the principles and methodologies that have guided the development of the DiMAT architecture, providing a detailed overview of its conceptual framework, the standards adopted, its modular and scalable design, and the strategies implemented to ensure efficient communication, secure data management and seamless integration with other systems. This approach aims to provide a robust technical solution and establish a replicable and adaptable architectural model for future developments in materials modelling and simulation. A. Architectural Framework and Standards The design of the DiMAT architecture is based on adopting principles and methodologies that ensure its viability, scalability, and adaptability over time. Implementing international standards has been a key factor in defining the architectural framework, ensuring that the developed solutions comply with best practices in software engineering and cyber-physical systems. By adhering to recognised standards and widely accepted reference architectures in industry and scientific research, the platform guarantees robust operation, interoperability with external systems, and the potential for future evolution without compromising its functionality. The DiMAT architecture has been developed following the ISO/IEC/IEEE 42010 standard, an international framework for system architectures' description, documentation, and analysis. This standard provides a structured methodological approach to defining system components, their relationships, and constraints, ensuring that each part of the architecture is developed with a clear purpose and in alignment with the system's technical requirements. Additionally, it facilitates communication among the various stakeholders involved in the platform's development and maintenance, enabling architectural decisions to be understood and validated by all relevant parties. Beyond this general regulatory framework, the design of DiMAT has been inspired by several reference architectures that have been successfully applied in other complex digital environments. Reviewing and analysing these architectures has allowed key elements to be extracted and incorporated into DiMAT’s structure to ensure efficiency, security, and flexibility. B. Industrial Internet Reference Architecture The Industrial Internet Reference Architecture (IIRA) [7] is a reference framework developed by the Industrial Internet Consortium (IIC) for implementing Industrial Internet systems. Its primary objective is to provide guidelines for integrating information technology and industrial automation, ensuring interoperability among devices, platforms, and systems in distributed environments. IIRA is structured around four primary viewpoints: • Business Viewpoint: Defines the architecture's strategic objectives and expected benefits, aligning technical requirements with business needs. • Usage Viewpoint: Describes the operational scenarios in which the system will interact with users and other systems. • Functional Viewpoint: Identifies key functional domains and their interactions, establishing the services and capabilities the architecture must provide. • Implementation Viewpoint: Details the specific technical aspects of the architecture, including technologies, communication protocols, and data models. In DiMAT, the structure of IIRA has been adopted to ensure seamless integration of the various toolkits and facilitate communication between them. The domain-based approach has enabled the design of an architecture where each toolkit operates autonomously within an interoperable ecosystem. C. Reference Architectural Model Industrie 4.0 The Reference Architectural Model Industry 4.0 (RAMI 4.0) [8] is a framework developed for digitalising industrial processes within the Industry 4.0 context. RAMI 4.0 provides a structured model based on three key dimensions: • Functional Layers: Defines six levels, from physical connectivity to data representation and decisionmaking based on processed information. • Lifecycle and Value Stream: Represents the evolution of industrial assets from their creation to decommissioning, ensuring information traceability throughout all production process stages. • Hierarchical Levels: Organises industrial assets into levels ranging from individual components to complete enterprise systems, allowing information integration at different scales. In the context of DiMAT, RAMI 4.0 has been a fundamental reference for structuring the architecture to effectively integrate the data generated in modelling, simulation, and manufacturing processes. The segmentation into layers has facilitated the definition of clear interfaces between the system’s components, ensuring efficient information flow across the various stages of the materials lifecycle. D. International Data Spaces Association Reference Architecture Model The International Data Spaces Association Reference Architecture Model (IDSA-RAM) [9] is a standard designed to facilitate secure data exchange in interconnected digital ecosystems. Its primary objective is ensuring that information can be shared between stakeholders without compromising data privacy, integrity, or intellectual property. IDSA-RAM is based on five primary layers: • Business Layer: Defines the roles and relationships among participants in the data ecosystem. • Functional Layer: Establishes the technical requirements necessary to ensure interoperability and access control to data. • Process Layer: Specifies data flows and governance rules for the secure exchange of information. • Information Layer: Defines formats and standards for structured data representation. • System Layer: Provides guidelines on implementing the necessary data management and storage technologies. In DiMAT, adopting the principles of IDSA-RAM has enabled the design of an infrastructure where information generated by the different toolkits can be shared securely and efficiently. Keycloak-based access control mechanisms have been implemented to ensure that only authorised users can access the system’s data and resources. E. Data Standardisation in Materials Modelling and Simulation Ensuring consistent, reusable, and machine-readable information is essential for advanced materials modelling and simulation. DiMAT integrates domain-specific standards that establish robust frameworks for describing, documenting, and sharing material data to address this requirement. Among the most relevant are EMMO (Elementary Mult perspective Material Ontology), MODA (Modelling Data), and CHADA (Characterisation Data). EMMO provides a structured top-level ontology for describing materials and their properties in a machinereadable format, facilitating the integration of experimental data and computational simulations. MODA establishes a framework for systematically documenting materials modelling processes, ensuring traceability and reproducibility of numerical simulations. CHADA, conversely, defines a standardised model for documenting materials characterisation methods, enabling experimental data to be interpreted and reused in different modelling and manufacturing contexts. In DiMAT, combining these standards has been key to ensuring that the data generated and processed within the platform can be efficiently shared among toolkits and external platforms. This facilitates DiMAT's integration into industrial and academic environments and promotes the adoption of best practices in data management within the materials modelling and simulation community. F. Modular and Scalable Design Principles The DiMAT architecture has been designed strongly emphasising modularity and scalability, allowing for its growth and adaptation without compromising system stability or performance. This approach addresses the need for a flexible platform capable of incorporating new functionalities efficiently without affecting the operation of existing modules. The platform consists of independent toolkits, each tailored to address specific end-user needs in digital transformation. These toolkits are organised into functional suites (for data management and assessment, modelling and design, and simulation and optimisation), which enable resource sharing and enhance interoperability between different components. Thanks to this modular structure, each toolkit can operate autonomously, facilitating its maintenance and updates without interfering with the rest of the platform. In terms of scalability, DiMAT has been designed to adapt to varying workloads through horizontal and vertical scaling strategies. Vertical scalability involves increasing resources on existing servers to enhance processing capacity, whereas horizontal scalability is based on replicating service instances to distribute the workload across multiple nodes. This latter approach ensures greater resilience and responsiveness to spikes in demand. Implementing technologies such as Docker and Kubernetes has been crucial in ensuring the platform’s scalability. Docker enables the containerisation of toolkits, providing consistent and portable execution environments. At the same time, Kubernetes facilitates the orchestration of these containers, optimising resource allocation and allowing for dynamic management based on demand. Additionally, load balancing mechanisms have been implemented to ensure that user requests are efficiently distributed among available instances, optimising resource utilisation and stable performance (Figure 1). Thanks to its modular and scalable design, DiMAT stands out as a robust, flexible, and future-ready platform. This architecture enables the efficient integration of new tools and ensures the optimal use of computational resources, facilitating its deployment across different environments and guaranteeing its long-term sustainability. G. Communication Architecture The DiMAT communication architecture has been designed to facilitate efficient, secure, and scalable interactions between toolkits, users, and external systems. Given the distributed nature of the platform, communication mechanisms must ensure seamless data exchange while maintaining low latency, high reliability, and strong security measures. To achieve this, DiMAT employs a hybrid communication model that integrates RESTful APIs and Message-Oriented Middleware (MOM), ensuring adaptability to different communication needs. DiMAT’s primary communication method relies on RESTful APIs over HTTPS, leveraging the ubiquity and efficiency of web-based communication. This approach allows for stateless interactions, where toolkits can request and retrieve information on demand, making the system lightweight and scalable. REST APIs ensure that toolkits, applications, and external services can interact through welldefined endpoints, following widely accepted industry standards for interoperability and ease of integration. Message-Oriented Middleware (MOM) is employed for more complex and asynchronous communication requirements. MOM provides publish-subscribe and point-topoint messaging models, allowing for decoupled service interactions. This approach is particularly beneficial for scenarios that require event-driven architectures, such as realtime monitoring, large-scale simulations, or high-frequency data exchanges. The RoboFuse platform supports traditional Message Queues (MQ) and modern RESTful API-based communication, ensuring flexibility and robustness in message delivery across toolkits. Beyond ensuring efficient data transmission, communication security is a critical aspect of the DiMAT architecture. All communication channels are encrypted using HTTPS/TLS protocols, guaranteeing the integrity and confidentiality of exchanged data. Firewalls and network segmentation strategies are implemented to prevent unauthorised access and malicious attacks, ensuring a multilayered defence model. The centralised authentication system, managed by Keycloak, plays a key role in securing communication, controlling user access, and verifying the legitimacy of API requests. Figure 2 illustrates how Keycloak is integrated into the DiMAT architecture, ensuring that authentication and authorisation are centrally managed across all toolkits and external services. DiMAT ensures that data flows efficiently between its toolkits and external stakeholders by integrating standardised, secure, and scalable communication mechanisms. This architecture allows for real-time interactions, seamless integration with third-party tools, and the ability to dynamically scale communication loads, reinforcing the platform’s adaptability and future-proofing infrastructure. H. Security and Identity Management Security and identity management are fundamental components of the DiMAT architecture, ensuring that all system interactions are protected, authenticated, and compliant with best cybersecurity practices. Given that the platform handles sensitive industrial and scientific data, it is essential to implement robust access control mechanisms, secure data storage policies, and continuous monitoring strategies to prevent unauthorised access or data breaches. The Identity and Access Management (IAM) system in DiMAT is built upon Keycloak. This widely used open-source identity provider enables Single Sign-On (SSO), multi-factor authentication (MFA), and role-based access control (RBAC). By leveraging Keycloak, DiMAT ensures that users only need to authenticate once to gain access to multiple toolkits while maintaining strict control over what data and functionalities they can access. The centralised authentication system simplifies user management across all toolkits and ensures a seamless and secure experience. To safeguard data integrity and confidentiality, DiMAT implements a layered security model that includes data encryption, network segmentation, and access monitoring. All sensitive data is stored using encrypted databases, ensuring that even if storage infrastructure is compromised, the information remains protected against unauthorised access. Furthermore, network security measures, such as firewalls, VPNs, and intrusion detection systems (IDS), are implemented to prevent external threats from accessing internal services. Fig. 1. Keycloak integration in the DiMAT Architecture Fig. 2. Docker horizontal scalability approach DiMAT employs a multi-tiered backup system that includes full, incremental, and differential backups as part of its disaster recovery and backup strategy. These backups are stored in secure, redundant locations to ensure data can be quickly restored in case of system failure or cyberattacks. Additionally, periodic penetration testing and security audits are conducted to identify and mitigate vulnerabilities, ensuring compliance with industry best practices and regulations such as GDPR. Another critical component of DiMAT’s security strategy is logging and monitoring. All system activities are recorded in structured logs, categorised by priority levels such as debug, info, warning, and error. These logs are analysed in real-time using automated anomaly detection algorithms, allowing security teams to proactively identify potential threats and mitigate risks before they escalate. By integrating state-of-the-art identity management, encryption, backup, and monitoring solutions, DiMAT provides a secure and resilient environment for data processing and simulation. This comprehensive security framework ensures that users can confidently interact with the platform, knowing their data and activities are protected against evolving cybersecurity threats. I. Interoperability and Standardization Interoperability is a core pillar of the DiMAT architecture, ensuring that toolkits and external systems can seamlessly exchange and process data. In an increasingly digital and interconnected industrial landscape, enabling smooth interaction between different software solutions, databases, and computational tools is essential for maximising efficiency and research potential. The interoperability model in DiMAT is designed to ensure compatibility, flexibility, and scalability, allowing users to integrate external resources while maintaining data integrity. To facilitate cross-platform compatibility, DiMAT employs APIs and semantic web technologies to unify data representation. RESTful APIs enable toolkits to communicate using a consistent protocol, ensuring that different applications can retrieve, process, and update data in a uniform manner. Additionally, the use of a common vocabulary (forming the basis for the DiMAT application ontology) along with semantic technologies in line with the FAIR principles facilitates interoperability [10]. The employed technologies, such as the Dataspace Management System (DSMS) and Neo4J, allow heterogeneous materials and process data to be harmonised and linked in a structured and meaningful way [11]. A structured methodology for interoperability has been established in DiMAT, ensuring that new toolkits and external applications can be integrated with minimal effort. This methodology involves: • Assessing interoperability requirements, ensuring that all components meet the necessary specifications. • Standardising communication protocols to facilitate seamless data exchange. • Employing modular design principles, ensuring that each toolkit operates independently while maintaining structured integration points. • Conducting rigorous integration testing, validating that new tools and datasets function correctly within the existing ecosystem. • Establishing feedback loops with users and developers to continuously improve integration mechanisms. DiMAT follows a structured methodology that involves requirements analysis, standard and protocol definition, and collaborative solution design to ensure seamless data exchange and interoperability across different toolkits and external platforms. This approach ensures that the integration of new technologies aligns with predefined interoperability standards while fostering collaboration among various partners. The interoperability process in DiMAT is driven by a combination of structured standardisation efforts and continuous monitoring, ensuring that all components operate cohesively within a unified framework. Furthermore, DiMAT promotes a collaborative ecosystem where stakeholders, developers, and researchers can contribute to and refine the platform’s interoperability mechanisms. By maintaining a commitment to open standards and transparent development practices, DiMAT ensures that it remains adaptable to emerging technologies and evolving industry needs. By integrating structured data representation, well-defined communication protocols, and a systematic approach to interoperability, DiMAT sets a new standard for seamless collaboration in materials modelling and simulation. This ensures that the platform can evolve without technical barriers, fostering innovation and facilitating crossdisciplinary research efforts. III. RESEARCH APPROACH The DiMAT Reference Architecture is designed as a structure that ensures scalability, security, interoperability, and adaptability to evolving technological demands. Following modular and distributed principles, this architecture provides a concrete model in which the various toolkits operate autonomously while maintaining seamless integration within a unified system. The architecture adopts a containerised microservices approach, where each component is encapsulated within a Docker-based service. This guarantees that toolkits remain independent, reducing software conflicts while facilitating efficient deployment and scalability. Each suite follows a service-oriented model, where applications communicate via standardised APIs, ensuring smooth data exchange between components. This structured approach promotes maintenance and the seamless integration of new toolkits or external applications without disrupting existing workflows. At the top level of the architecture, an authentication layer centralises user access management and security enforcement. This ensures that all interactions within the platform remain secure, with role-based access control and centralised authentication mechanisms in place. Additionally, the architecture includes a frontend interface layer, which provides an access point for each toolkit, facilitating interaction with the system's functionalities. This layer guarantees a consistent user experience, simplifying access to modelling, simulation, and data management services. Another key feature of the DiMAT Reference Architecture is its communication and interoperability model, which ensures that toolkits can efficiently share and process data. To achieve this, standardised API interfaces have been implemented, enabling structured communication between the various components while preventing compatibility issues. The reference architecture is deployed within a cloudbased environment, structured into three primary layers: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). The IaaS layer represents the cloud computing infrastructure where resources are provisioned and managed. The PaaS layer consists of the software execution environment, where containerised services run within orchestrated cloud infrastructures. Finally, the SaaS layer includes web-based applications, allowing researchers and engineers to interact with the system through browseraccessible interfaces. This cloud-native approach ensures high availability, scalability, and efficient management of computational resources. Figure 4 provides a graphical representation of the DiMAT Reference Architecture, illustrating the three functional suites, their interconnected services, and the cloud-based deployment model on which the platform operates. This image visually demonstrates how toolkits interact within the ecosystem, ensuring an efficient, scalable, and secure environment for advanced materials research and simulation. IV. FINDINGS A. Containerisation with Docker The DiMAT architecture adopts containerisation as the foundation for software deployment, ensuring portability, consistency, and scalability across different environments. The adoption of Docker, one of the most widely used containerisation platforms, allows developers to package applications and their dependencies into lightweight, selfsufficient units that can be easily deployed and executed across various infrastructures. This methodology not only simplifies the development workflow but also enhances resource efficiency and system reliability. B. The Role of Docker in the DiMAT Ecosystem Docker serves as the core of the software development lifecycle in DiMAT, enabling toolkits and applications to be packaged as standardised, portable containers. Each Docker container encapsulates the application code, execution environment, libraries, and dependencies required for operation, ensuring that applications run identically across all environments, regardless of the underlying system configuration. One of the key benefits of Docker in DiMAT is its ability to eliminate compatibility issues. Traditionally, software deployment is affected by discrepancies between development, testing, and production environments. However, by encapsulating applications within Docker containers, the DiMAT project ensures that software behaves consistently across all deployment stages, reducing the risk of unexpected failures caused by system configuration differences. Additionally, Docker enables efficient resource execution by using containerised microservices rather than deploying full virtual machines. This approach results in faster startup times, reduced overhead, and improved scalability, making it ideal for high-performance scientific computing tasks, such as materials modelling, numerical simulations, and data processing within DiMAT. C. Containerised Architecture in DiMAT The Docker-based architecture within DiMAT follows a modular, service-oriented design, where different toolkits and platform services are packaged as independent containers. These containers are managed using Docker Compose and orchestrated via Kubernetes, ensuring smooth execution and coordination across distributed and cloud environments. The containerisation workflow in DiMAT consists of the following key steps: 1. Container Definition and Image Creation: Each software component is encapsulated within a Docker image, defining the necessary runtime environment, dependencies, and configurations. This image serves as a standardised template for container execution. 2. Version Control and Registry Management: The generated Docker images are stored in the DiMAT registry, providing a secure and centralised repository for managing different software versions. 3. Automated Deployment and Execution: Once an image is ready, it is deployed to designated execution environments, either in local test environments or within DiMAT's cloud infrastructure, for large-scale simulations and computations. 4. Container Orchestration: Kubernetes manages multicontainer deployments, automatically scaling workloads, balancing resources, and handling failures. This containerised architecture ensures that each component operates within a well-defined, isolated environment, minimising dependency conflicts and providing a stable foundation for continuous software delivery. D. Future Enhancements in DiMAT’s Containerisation Strategy As the DiMAT platform continues to evolve, additional optimisations in containerisation are being explored to further enhance performance, security, and maintainability. One major area of focus is the optimisation of container images, which involves reducing image size by eliminating unnecessary dependencies. This strategy speeds up software deployment and lowers overall resource consumption. Another key initiative is the adoption of AI-driven container orchestration, where machine learning algorithms are utilised to dynamically adjust resource allocation based on workload patterns, thereby improving system efficiency and responsiveness. In parallel, enhanced security protocols are being integrated, including automated vulnerability scanning within the Docker build phase to proactively detect and mitigate potential threats. Additionally, validation benchmarks are planned to systematically assess performance, reliability, and compliance across containerised deployments. By continuously refining its containerisation strategy, DiMAT ensures that its software solutions remain flexible, scalable, and resilient, supporting advancements in materials modelling, simulations, and scientific computing. E. Continuous Integration and Continuous Deployment (CI/CD) Contemporary software engineering practices emphasise automation and efficiency to facilitate the prompt deployment of dependable applications. Continuous integration and deployment have emerged as foundational methodologies that streamline the development process and diminish errors while expediting software delivery [12]. The DiMAT project incorporates CI/CD methodologies to automate software testing and deployment, ensuring that each solution update is swiftly validated and seamlessly transitioned into a production environment. Implementing this approach not only enhances the efficiency of development workflows but also guarantees consistency and reliability in the software release cycle. F. The CI/CD Framework in DiMAT The CI/CD framework within the DiMAT project is designed to facilitate a structured and automated software development pipeline. This pipeline follows a three-phase approach, including building, shipping, and running containerised applications. The process begins in the development phase, where developers write and modify solutions using preferred environments. Once changes are made, the updated code is pushed to GitLab, which serves as the central version control system. This integration triggers automated processes that install necessary dependencies and execute predefined unit tests, ensuring that any modifications adhere to project requirements before proceeding further. After successfully passing the testing phase, the code transitions to the shipping stage, where a Docker image is generated. This image packages the software and all its required dependencies, ensuring platform independence and consistent execution across various environments. The produced Docker image is then uploaded to a secure registry, the DiMAT registry, for later deployment. The use of containerisation in this phase guarantees the isolation and reproducibility of each software build, mitigating potential compatibility problems[13]. The final stage involves deploying the Docker image from the registry to a designated execution environment. The deployment phase commences after successful shipping, where the validated software images are transferred to cloud environments, specifically to the DiMAT cloud infrastructure. This deployment is managed through GitLab CI/CD solutions, which oversee the automation of updates and facilitate the seamless transition from development to live production. The newly deployed software undergoes additional validation by testers and users before being fully integrated into the production environment. This final stage ensures that the software remains stable and functions as intended, thereby maintaining the system's integrity. In large solution projects, the accuracy, reliability, and availability of the Software Release Management process are critical, encompassing the planning, design, build, configuration, testing, and deployment of associated software components [14]. Such deployment pipelines consist of multiple stages, where each stage is triggered only when the preceding stage has been successfully completed, thereby ensuring a smooth transition from continuous integration to continuous delivery [15] (Figure 4). G. The Developer’s Perspective in CI/CD From a developer's perspective, the Continuous Integration/Continuous Deployment (CI/CD) workflow within the DiMAT project is structured to provide a systematic and efficient approach to software deployment. The process commences when a developer submits updated code to GitLab, which subsequently initiates an automated job responsible for installing dependencies and executing essential tests. If the code meets the requisite validation criteria, a Docker image is constructed and subsequently transferred to the registry. Once the image is available in the registry, it is deployed into a staging container, where additional validation and testing are conducted. The subsequent phase involves upgrading the production container, which in the DiMAT framework necessitates manual approval. This manual intervention serves as a safeguard, enabling teams to verify the software's performance in the staging environment before promoting it to production. While the deployment into staging is fully automated, this final step ensures that potential risks are mitigated before the software is made available to end-users. H. Advantages of CI/CD Implementation in DiMAT Incorporating Continuous Integration and Continuous Deployment (CI/CD) in the DiMAT initiative confers advantages that enhance software development and deployment processes. A primary benefit involves automated procedures, which mitigate the necessity of human oversight for the build, deployment, and execution of toolkits. By automating these tasks, practitioners can focus on refining functionality rather than overseeing infrastructural elements. This structural framework also ensures a standardised environment, given that Docker images encapsulate all essential dependencies, thus eliminating potential conflicts arising from divergent configurations. Another significant advantage emerges from the early detection of defects through automated testing. By continuously integrating and evaluating new code, the system uncovers potential issues initially, preventing their escalation into complications in live environments. Such a preemptive methodology promotes enhanced software resilience and reliability. Furthermore, utilising cloud-based platforms facilitates scalability, ensuring numerous deployments function across distributed infrastructures. Through a welldesigned CI/CD pipeline, the DiMAT project achieves accelerated iteration cycles, enabling frequent software refinements without inducing disruptions or service interruptions. Despite the expected benefits of the CI/CD approach, comprehensive quantitative metrics on improvements in deployment times, error reduction or scalability benchmarks have not currently been collected. Such assessments are planned for future phases of the project, once the integration of all toolkits has been completed and a consolidated operational environment is in place. Nevertheless, the adoption of CI/CD has been well received by technical users within the DiMAT consortium, who have reported a marked improvement in the smoothness of the development and deployment process. These impressions are based on preliminary observations in test environments, where automation has facilitated the continuous integration of new functionalities and the early detection of bugs. Some challenges have been identified related to the dependency on specific technologies such as Docker and Kubernetes, which may represent a barrier for teams with less experience in cloud-native infrastructures. To address this, training materials and technical documentation are being developed to support progressive adoption in less digitally mature contexts. The CI/CD pipeline adopted within the DiMAT project exemplifies the efficacy of contemporary DevOps approaches in streamlining development workflows. By leveraging GitLab CI/CD, Docker, and cloud-based deployment mechanisms, the project attains a high level of automation, ensuring that software revisions are verified and launched. This methodology considerably reduces the interval required to migrate from development to production while enhancing software robustness and consistency concurrently. Although originally designed for the domain of materials modelling, the DiMAT architecture presents a modular and secure structure that is generalisable to other industrial contexts. Potential applications include predictive maintenance, supply chain optimisation, and even biomedical simulations, where containerised deployment and data interoperability are equally critical. To further validate the platform’s usability and impact, future work will include use case evaluations from participating SMEs and academic institutions. These case studies will help illustrate real-world deployment experiences and sector-specific benefits. One distinctive feature of DiMAT’s architecture is the integration of domain-specific ontologies (such as EMMO, MODA, and CHADA) directly within the CI/CD pipeline. This enables not only semantic consistency but also traceable, machinereadable workflows—an uncommon yet powerful approach in current DevOps implementations. The DiMAT project aims to refine its CI/CD operations by incorporating advanced validation methods, including AIdriven testing strategies, to further enhance automation potential. Moreover, integrating augmented security protocols into the pipeline will better safeguard the system from vulnerabilities and unauthorised modifications. Incorporating CI/CD into the DiMAT project exemplifies current cloudbased development practices, effectively demonstrating how automated integration and deployment can enhance efficiency, reliability, and innovation within large-scale research and development endeavours. V. CONCLUSIONS The DiMAT architecture represents a significant step forward in addressing the complexities and demands of modern industrial ecosystems, particularly in materials and process data management, modelling, simulation, and design. By embracing industry-leading standards and frameworks like ISO/IEC/IEEE 42010, IIRA, RAMI 4.0, and IDSA-RAM, DiMAT achieves a balance between modularity, scalability, and robust security. The containerised microservices design, powered by Docker and Kubernetes, ensures that the platform can adapt to fluctuating computational and functional requirements without sacrificing performance or reliability. Moreover, integrating comprehensive security protocols and Keycloak-based identity management underscores DiMAT’s commitment to safeguarding sensitive data across all research and production lifecycle stages. This layered security model and established backup and monitoring strategies provide resilience against system failures and cybersecurity threats. The adherence to interoperability standards such as EMMO, MODA, and CHADA further enhances the platform’s versatility, enabling seamless data exchange and collaboration with external tools and services. From a development perspective, implementing CI/CD methodologies has proven essential in enabling frequent and reliable software updates. DiMAT ensures a streamlined development process that can rapidly incorporate new functionalities by automating testing, deployment, and version control. The platform’s design remains open to future enhancements, such as AI-driven container orchestration and advanced security analysis, to meet the evolving needs of Industry 4.0 and scientific research. Ultimately, DiMAT lays a robust foundation for innovation in materials modelling, creating a scalable, secure, and interoperable environment ready to address emerging challenges in industrial digital transformation. ACKNOWLEDGEMENT This paper was funded by the European Union’s Horizon Europe research and innovation programme under grant agreement No. 101091496, project DiMAT (Digital Modelling and Simulation for Design, Processing and Manufacturing of Advanced Materials). REFERENCES [1] K. Y. Akdil, A. Ustundag, and E. Cevikcan, Industry 4.0: Managing The Digital Transformation. 2018. [Online]. Available: https://doi.org/10.1007/978-3-319-57870-5_4 [2] K. Tsitseklis et al., “A Reference Architecture for Digital Transformation of SMEs in the Manufacturing Domain,” 2023. [3] ISO.ORG, “ISO/IEC/IEEE 42010:2011 Systems and software engineering — Architecture description,” 2011. [Online]. Available: https://www.iso.org/standard/50508.html [4] K. Aruna and P. Gurunathan, “Enhancing Edge Environment Scalability: Leveraging Kubernetes for Container Orchestration and Optimization,” Concurr Comput, pp. 1–11, 2024, doi: 10.1002/cpe.8303. [5] Oyekunle Claudius Oyeniran, Oluwole Temidayo Modupe, Aanuoluwapo Ayodeji Otitoola, Oluwatosin Oluwatimileyin Abiona, Adebunmi Okechukwu Adewusi, and Oluwatayo Jacob Oladapo, “A comprehensive review of leveraging cloud-native technologies for scalability and resilience in software development,” International Journal of Science and Research Archive, vol. 11, no. 2, pp. 330–337, 2024, doi: 10.30574/ijsra.2024.11.2.0432. [6] A. Chatterjee, M. W. Gerdes, P. Khatiwada, and A. Prinz, “Applying Spring Security Framework with TSD-Based OAuth2 to Protect Microservice Architecture APIs,” IEEE Access, vol. 10, pp. 41900–41920, 2022, doi: 10.1109/ACCESS.2022.3165548. [7] S.-W. Lin et al., “The Industrial Internet of Things Volume G1 : Reference Architecture,” Industrial Internet Consortium White Paper, vol. Version 1., p. 58 Seiten, 2017. [8] G. Koschnick, M. Hankel, and B. Rexroth, “The Reference Architectural Model Industrie 4.0 (RAMI 4.0),” 2015. [Online]. Available: www.zvei.org/ [9] International Data Spaces Association, “International Data Spaces - Reference Architecture Model,” https://docs.internationaldataspaces.org/ids-knowledgebase/idsram-4, 2022. [10] M. D. Wilkinson et al., “The FAIR Guiding Principles for scientific data management and stewardship,” Sci Data, vol. 3, pp. 1–9, 2016, doi: 10.1038/sdata.2016.18. [11] Y. Nahshon et al., “Semantic Orchestration and Exploitation of Material Data: A Dataspace Solution Demonstrated on Steel and Copper Applications,” Adv Eng Mater, vol. 2401448, 2024, doi: 10.1002/adem.202401448. [12] V. K. Thatikonda, “Beyond the Buzz: A Journey Through CI/CD Principles and Best Practices,” European Journal of Theoretical and Applied Sciences, vol. 1, no. 5, pp. 334–340, 2023, doi: 10.59324/ejtas.2023.1(5).24. [13] M. Canesche, R. Leißa, and F. M. Q. Pereira, “Preparing Reproducible Scientific Artifacts using Docker,” 2023, Zenodo. doi: 10.5281/zenodo.8096830. [14] M. C. S. Fhang and R. Swamy, “Best practices in release management of large projects,” ACM International Conference Proceeding Series, pp. 51–55, 2018, doi: 10.1145/3185089.3185112. [15] S. Vöst and S. Wagner, “Towards Continuous Integration and Continuous Delivery in the Automotive Industry,” 2016, [Online]. Available: http://arxiv.org/abs/1612.04139