Cloud Infrastructure Tools For Cloud Applications : Infrastructure management of multiple cloud platforms
Valkeinen, Matti (2022)
2022
Teknis-luonnontieteellinen DI-ohjelma - Master's Programme in Science and Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2022-03-08
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202202101909
https://urn.fi/URN:NBN:fi:tuni-202202101909
Tiivistelmä
Cloud platforms are the modern solution to host applications as a SaaS. The cloud platforms share standard practices, but the services as simple as cloud storage differ significantly between each provider. This master's thesis goes through a case study about Profit Software's application hosted in the AWS cloud environment, aiming to transfer to Azure and Google Cloud.
The transfer of already existing cloud applications requires identifying the key services on the original cloud platform and planning the appropriate counterpart in another cloud platform. When cloud architecture is designed for desired cloud platforms comes the question of how one implements them to keep the infrastructure in check. The use of IaC tools helps the setup of cloud infrastructure and maintenance while providing consistent results during the development process and when moving to production.
The IaC tools considered for the Profit Software's application were Terraform, Pulumi and cloud providers' AWS CDK, ARM and Google Cloud Deployment Manager. And since the application uses AWS ECS as its container orchestrators, a new orchestrator had to be also evaluated. Considered orchestrators were Docker Swarm, Kubernetes and each cloud platforms' orchestrator.
The first cloud transferability evaluation was done against the Profit Software's application by checking if the two other clouds have the necessary services to check if they had the required services to run the application. The use of AWS Batch presented a problem since Google Cloud doesn't yet have a corresponding service. A similar service is a Google Cloud Run, but their service logic differs so that the application requires some modifications before the transfer. On the other hand, Azure had the necessary services, and they function similarly enough to their AWS counterparts.
The IaC and container orchestrators were evaluated with help on a smaller project with key features similar to the Profit's application. The various IaC tools were tested how they would suffice to build the infrastructure for this smaller project. The rest of the services that weren't present in the project were checked through IaC tools' documentation if it would be possible to create them. The best of the considered IaC tools were Pulumi and Terraform, but Pulumi takes edge over the Terraform with the flexibility of its way of infrastructure defining compared to Terraform. The considered container orchestrators were all enough for the smaller project and the actual application. But when considering the benefits they provide, Kubernetes took the lead by far since all of the planned cloud platforms support Kubernetes clusters.
The transfer of already existing cloud applications requires identifying the key services on the original cloud platform and planning the appropriate counterpart in another cloud platform. When cloud architecture is designed for desired cloud platforms comes the question of how one implements them to keep the infrastructure in check. The use of IaC tools helps the setup of cloud infrastructure and maintenance while providing consistent results during the development process and when moving to production.
The IaC tools considered for the Profit Software's application were Terraform, Pulumi and cloud providers' AWS CDK, ARM and Google Cloud Deployment Manager. And since the application uses AWS ECS as its container orchestrators, a new orchestrator had to be also evaluated. Considered orchestrators were Docker Swarm, Kubernetes and each cloud platforms' orchestrator.
The first cloud transferability evaluation was done against the Profit Software's application by checking if the two other clouds have the necessary services to check if they had the required services to run the application. The use of AWS Batch presented a problem since Google Cloud doesn't yet have a corresponding service. A similar service is a Google Cloud Run, but their service logic differs so that the application requires some modifications before the transfer. On the other hand, Azure had the necessary services, and they function similarly enough to their AWS counterparts.
The IaC and container orchestrators were evaluated with help on a smaller project with key features similar to the Profit's application. The various IaC tools were tested how they would suffice to build the infrastructure for this smaller project. The rest of the services that weren't present in the project were checked through IaC tools' documentation if it would be possible to create them. The best of the considered IaC tools were Pulumi and Terraform, but Pulumi takes edge over the Terraform with the flexibility of its way of infrastructure defining compared to Terraform. The considered container orchestrators were all enough for the smaller project and the actual application. But when considering the benefits they provide, Kubernetes took the lead by far since all of the planned cloud platforms support Kubernetes clusters.