Open AccessMLOps approach in the cloud-native data pipeline design
Author(s) -
István Pölöskei
Publication year - 2021
Publication title -
acta technica jaurinensis
Language(s) - English
Resource type - Journals
eISSN - 2064-5228
pISSN - 1789-6932
DOI - 10.14513/actatechjaur.00581
Subject(s) - cloud computing , workflow , pipeline (software) , computer science , software deployment , data science , big data , context (archaeology) , analytics , process (computing) , software engineering , database , data mining , paleontology , biology , programming language , operating system
The data modeling process is challenging and involves hypotheses and trials. In the industry, a workflow has been constructed around data modeling. The offered modernized workflow expects to use of the cloud’s full abilities as cloud-native services. For a flourishing big data project, the organization should have analytics and information-technological know-how. MLOps approach concentrates on the modeling, eliminating the personnel and technology gap in the deployment. In this article, the paradigm will be verified with a case-study in the context of composing a data pipeline in the cloud-native ecosystem. Based on the analysis, the considered strategy is the recommended way for data pipeline design.