Migrating Python Application to AWS Elastic Container Service
Takkunen, Juhani (2021)
2021
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021082317090
https://urn.fi/URN:NBN:fi:amk-2021082317090
Tiivistelmä
An accounting company Azets Insight Oy had developed a custom integration platform, written in Python programming language, which was running on multiple self-hosted Windows and Linux machines. While Python virtual environments can work in both operating systems, some libraries and commands failed to work on every machine due to different operating systems or versioning, for example. The malfunctions resulted in unexpected application behaviour and crashes both in the development and production environments.
Since the company also had a strategic goal to utilize Amazon Web Services (AWS), this thesis investigated how an application could be containerized and migrated to AWS Elastic Container Service. Containers were selected because they allow developers to test applications on their local machines and to expect the applications to behave similarly in the production system. On the other hand, AWS Elastic Container Service was considered as a scalable Platform as a Service for running the containers, thus reducing manual upkeep.
A literature overview was conducted for containers and container-related tools offered by Cloud Service Providers in general, with special emphasis on AWS tools due to company interest. Empirical participant observations were done in the Spring of 2020 by the author, resulting in the current state analysis.
The integration application's current solutions were compared to Amazon Web Services tools, such as Fargate, CloudWatch, ElasticSearch, S3, and RDS. Based on the findings, a guideline was created for migrating the integration platform to AWS. The guideline recommends that the implementation is done in three phases to allow developers to focus on fewer tasks at the time.
This study showed that running the integration application in AWS Elastic Container Service is a viable option and it can reduce manual maintenance compared to the current state. AWS can also offer tooling that is not currently available. The study did not attempt to estimate or compare costs for implementing nor operating applications in AWS, although the findings would be interesting if such a study was later performed.
While the focus of the thesis was to explore how containers and AWS could be used within a specific company, the results and especially the reasoning should be interesting for anyone considering migrating applications to a container service.
Since the company also had a strategic goal to utilize Amazon Web Services (AWS), this thesis investigated how an application could be containerized and migrated to AWS Elastic Container Service. Containers were selected because they allow developers to test applications on their local machines and to expect the applications to behave similarly in the production system. On the other hand, AWS Elastic Container Service was considered as a scalable Platform as a Service for running the containers, thus reducing manual upkeep.
A literature overview was conducted for containers and container-related tools offered by Cloud Service Providers in general, with special emphasis on AWS tools due to company interest. Empirical participant observations were done in the Spring of 2020 by the author, resulting in the current state analysis.
The integration application's current solutions were compared to Amazon Web Services tools, such as Fargate, CloudWatch, ElasticSearch, S3, and RDS. Based on the findings, a guideline was created for migrating the integration platform to AWS. The guideline recommends that the implementation is done in three phases to allow developers to focus on fewer tasks at the time.
This study showed that running the integration application in AWS Elastic Container Service is a viable option and it can reduce manual maintenance compared to the current state. AWS can also offer tooling that is not currently available. The study did not attempt to estimate or compare costs for implementing nor operating applications in AWS, although the findings would be interesting if such a study was later performed.
While the focus of the thesis was to explore how containers and AWS could be used within a specific company, the results and especially the reasoning should be interesting for anyone considering migrating applications to a container service.