Analysis and Development of the MLOps Pipeline for ML Model Deployment
Main Article Content
Abstract
The growth in the number of IT products with machine-learning features is increasing the relevance of automating machine-learning processes. The use of MLOps techniques is aimed at providing training and efficient deployment of applications in a production environment by automating side infrastructure issues that are not directly related to model development.
In this paper, we review the components, principles, and approaches of MLOps and analyze existing platforms and solutions for building machine learning pipelines. In addition, we propose an approach to build a machine learning pipeline based on basic DevOps tools and open-source libraries.
Article Details
References
2. Van der Meulen R., McCall T. Gartner Says Nearly Half of CIOs Are Planning to Deploy Artificial Intelligence // Gartner. 2018. URL: https://www.gartner.com/en/newsroom/press-releases/2018-02-13-gartner-says-nearly-half-of-cios-are-planning-to-deploy-artificial-intelligence (дата обращения: 01.06.2022).
3. Posoldova A. Machine Learning Pipelines: From Research to Production // IEEE Potentials. 2020. Vol. 39, No. 6. P. 38–42. https://doi.org/10.1109/MPOT.2020.3016280
4. Alla S., Adari S.K. What is mlops? // In: Beginning MLOps with MLFlow. Berkeley: Apress, 2021. P. 79–124.
5. Gift N., Deza A. Practical MLOps. O'Reilly Media, Inc., 2021.
6. Symeonidis G., Nerantzis E., Kazakis A., Papakostas G.A. MLOps – Definitions, Tools and Challenges // IEEE Annual Computing and Communication Workshop and Conference (CCWC). Las Vegas. 2022. Vol. 12. P. 453–460. https://doi.org/10.48550/arXiv.2201.00162
7. Chen L. Continuous Delivery: Huge Benefits, but Challenges Too // IEEE Software. 2015. Vol. 32. P. 50–54. https://doi.org/10.1109/MS.2015.27
8. John M., Olsson H., Bosch J. Towards MLOps: A Framework and Maturity Model // Euromicro Conference on Software Engineering and Advanced Applications (SEAA). Palermo. 2021. Vol. 47. P. 1–8. https://doi.org/10.1109/SEAA53835.2021.00050
9. MLOps: Continuous delivery and automation pipelines in machine learning // Google Cloud. 2021. URL: https://cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning (дата обращения: 03.07.2021).
10. Machine Learning operations maturity model // Microsoft. URL: https://docs.microsoft.com/en-us/azure/architecture/example-scenario/mlops/ mlops-maturity-model, last accessed 2022/05/30.
11. Kreuzberger D., Kühl N., Hirschl S. Machine Learning Operations (MLOps): Overview, Definition, and Architecture // arXiv preprint arXiv:2205.02302. 2022. https://doi.org/10.48550/arXiv.2205.02302
12. MLOps Principles // MLOps. URL: https://ml-ops.org/content/mlops-principles (дата обращения: 03.07.2021).
13. Yandex DataSphere // Yandex Cloud. URL: https://cloud.yandex.ru/services/datasphere, (дата обращения: 05.03.2022).
14. Проект // Yandex datasphere документация. URL: https://cloud.yandex.ru/docs/datasphere/concepts/project, (дата обращения: 05.03.2022).
15. Развертывание эксплуатации моделей // Yandex datasphere документация. URL: https://cloud.yandex.ru/docs/datasphere/concepts/deploy, (дата обращения: 05.03.2022).
16. MLFlow. URL: https://mlflow.org (дата обращения: 28.12.2021).
17. MLflow Tracking // MLflow. URL: https://mlflow.org/docs/latest/tracking.html (дата обращения: 28.12.2021).
18. MLflow Projects // MLflow. URL: https://mlflow.org/docs/latest/projects.html, (дата обращения: 28.12.2021).
19. MLflow Models // MLflow. URL: https://mlflow.org/docs/latest/models.html (дата обращения: 28.12.2021).
20. MLflow Model Registry // MLflow. URL: https://mlflow.org/docs/latest/model-registry.html (дата обращения: 28.12.2021).
21. Khandelwal N. MLflow Alternatives for Data Version Control: DVC vs. MLflow // Censious. URL: https://censius.ai/blogs/dvc-vs-mlflow (дата обращения: 30.05.2022).
22. Hewage N., Meedeniya D. Machine Learning Operations: A Survey on MLOps Tool Support // arXiv preprint arXiv:2202.10169. 2022. https://doi.org/10.48550/arXiv.2202.10169
23. Introduction // Kubeflow documentation. URL: https://www.kubeflow.org/docs/started/introduction (дата обращения: 11.03.2022).
24. What is Kubeflow? // Kubeflow. URL: https://www.kubeflow.org (дата обращения: 11.03.2022).
25. Architecture // Kubeflow documentation. URL: https://www.kubeflow.org/docs/started/architecture (дата обращения: 11.03.2022).
26. Kaewsanmua K. Best 8 Machine Learning Model Deployment Tools That You Need to Know // Neptune. 2021. URL: https://neptune.ai/blog/best-8-machine-learning-model-deployment-tools (дата обращения: 01.06.2022).
27. DVC. URL: https://dvc.org (дата обращения: 27.12.2021).
28. Zhao Y. MLOps: Data versioning with DVC — Part Ⅰ // Medium. 2020. URL: https://yizhenzhao.medium.com/mlops-data-versioning-with-dvc-part-ⅰ-8b3221df8592 (дата обращения: 27.12.2021).
29. Mesquita D. The ultimate guide to building maintainable Machine Learning pipelines using DVC // Towards data science. 2020. URL: https://towardsdatascience.com/the-ultimate-guide-to-building-maintainable-machine-learning-pipelines-using-dvc-a976907b2a1b (дата обращения: 27.12.2021).
30. CML Documentation // CML. URL: https://cml.dev/doc (дата обращения: 27.12.2021).
31. Continuous Integration and Deployment for Machine Learning // DVC. URL: https://dvc.org/doc/use-cases/ci-cd-for-machine-learning (дата обращения: 27.12.2021).
32. Continuous Integration with CML and Github Actions // MLOps Guide. URL: https://mlops-guide.github.io/CICD/cml_testing (дата обращения: 27.12.2021).
33. Kubernetes Documentation // Kubernetes. URL: https://kubernetes.io/docs/hom (дата обращения: 22.05.2022).
34. What is Prometheus? // Prometheus. URL: https://prometheus.io/docs/introduction/overview
35. Grafana // Grafana Labs. URL: https://grafana.com/grafana (дата обращения: 22.05.2022).
This work is licensed under a Creative Commons Attribution 4.0 International License.
Presenting an article for publication in the Russian Digital Libraries Journal (RDLJ), the authors automatically give consent to grant a limited license to use the materials of the Kazan (Volga) Federal University (KFU) (of course, only if the article is accepted for publication). This means that KFU has the right to publish an article in the next issue of the journal (on the website or in printed form), as well as to reprint this article in the archives of RDLJ CDs or to include in a particular information system or database, produced by KFU.
All copyrighted materials are placed in RDLJ with the consent of the authors. In the event that any of the authors have objected to its publication of materials on this site, the material can be removed, subject to notification to the Editor in writing.
Documents published in RDLJ are protected by copyright and all rights are reserved by the authors. Authors independently monitor compliance with their rights to reproduce or translate their papers published in the journal. If the material is published in RDLJ, reprinted with permission by another publisher or translated into another language, a reference to the original publication.
By submitting an article for publication in RDLJ, authors should take into account that the publication on the Internet, on the one hand, provide unique opportunities for access to their content, but on the other hand, are a new form of information exchange in the global information society where authors and publishers is not always provided with protection against unauthorized copying or other use of materials protected by copyright.
RDLJ is copyrighted. When using materials from the log must indicate the URL: index.phtml page = elbib / rus / journal?. Any change, addition or editing of the author's text are not allowed. Copying individual fragments of articles from the journal is allowed for distribute, remix, adapt, and build upon article, even commercially, as long as they credit that article for the original creation.
Request for the right to reproduce or use any of the materials published in RDLJ should be addressed to the Editor-in-Chief A.M. Elizarov at the following address: amelizarov@gmail.com.
The publishers of RDLJ is not responsible for the view, set out in the published opinion articles.
We suggest the authors of articles downloaded from this page, sign it and send it to the journal publisher's address by e-mail scan copyright agreements on the transfer of non-exclusive rights to use the work.