This report describes the internship I carried out at MYS S.r.l. in Rovigo from 15 June to 7 August 2026, for a total of 300 hours. The goal was to migrate an institutional web service from virtual machines to containers on the Amazon Web Services platform, and to automate the process by which a new version is deployed. Before the work, a release required seven manual steps that only one person in the company was able to perform. The system describes the infrastructure with Terraform, declaratively and under version control, runs the service in containers on Amazon ECS and AWS Fargate, and delegates the release to a GitHub Actions pipeline that builds the image, publishes it to Amazon ECR, and hands the new version over to AWS CodeDeploy. The latter deploys the new version alongside the previous one, shifts traffic only once health checks pass, and shifts it back when Amazon CloudWatch alarms detect a faulty release. On the pilot service, manual steps go from seven to one, the time from publishing a version to a completed release drops from about one hour to just over eight minutes, and rollback to the previous version happens without human intervention.
La relazione descrive il tirocinio che ho svolto presso MYS S.r.l. di Rovigo dal 15 giugno al 7 agosto 2026, per una durata complessiva di 300 ore. L'obiettivo era migrare un servizio web istituzionale da macchine virtuali a container sulla piattaforma Amazon Web Services e automatizzare il processo con cui una nuova versione entra in esercizio. Prima dell'intervento il rilascio richiedeva sette operazioni manuali, che una sola persona in azienda era in grado di condurre. Il sistema realizzato descrive l'infrastruttura con Terraform, in forma dichiarativa e versionata, esegue il servizio in container su Amazon ECS e AWS Fargate, e affida il rilascio a una catena GitHub Actions che costruisce l'immagine, la pubblica su Amazon ECR e consegna la versione nuova ad AWS CodeDeploy. Quest'ultimo affianca la versione nuova alla precedente, sposta il traffico soltanto quando i controlli di integrità hanno esito positivo e lo riporta indietro quando gli allarmi di Amazon CloudWatch rilevano un rilascio difettoso. Sul servizio scelto come pilota le operazioni manuali passano da sette a una, il tempo dalla pubblicazione di una versione al rilascio compiuto scende da circa un'ora a poco più di otto minuti, e il ritorno alla versione precedente avviene senza intervento umano.
Migrazione architetturale e automazione del rilascio applicativo
EZZEMOURI, HOSSAM
2025/2026
Abstract
This report describes the internship I carried out at MYS S.r.l. in Rovigo from 15 June to 7 August 2026, for a total of 300 hours. The goal was to migrate an institutional web service from virtual machines to containers on the Amazon Web Services platform, and to automate the process by which a new version is deployed. Before the work, a release required seven manual steps that only one person in the company was able to perform. The system describes the infrastructure with Terraform, declaratively and under version control, runs the service in containers on Amazon ECS and AWS Fargate, and delegates the release to a GitHub Actions pipeline that builds the image, publishes it to Amazon ECR, and hands the new version over to AWS CodeDeploy. The latter deploys the new version alongside the previous one, shifts traffic only once health checks pass, and shifts it back when Amazon CloudWatch alarms detect a faulty release. On the pilot service, manual steps go from seven to one, the time from publishing a version to a completed release drops from about one hour to just over eight minutes, and rollback to the previous version happens without human intervention.| File | Dimensione | Formato | |
|---|---|---|---|
|
Ezzemouri_Hossam.pdf
Accesso riservato
Dimensione
1.3 MB
Formato
Adobe PDF
|
1.3 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/116486