Hybrid disaster-recovery solution using public cloud

Project goal

In 2020, the Database Services group in the CERN IT department launched this project, in collaboration with Oracle, to explore how integration of commercial cloud platforms with CERN on-premises systems might improve the resilience of services. The aim of this project is to understand the benefits, limitations and cost of the cloud environment for application servers and databases.

R&D topic
Data-centre technologies and infrastructures
Project coordinator(s)
Viktor Kozlovzky
Team members
Aimilios Tsouvelekakis, Alina Andreea Grigore, Andrei Dumitru, Antonio Nappi, Arash Khodabandeh, Artur Wiecek, Borja Aparicio Cotarelo, Edoardo Martelli, Ignacio Coterillo Coz, Sebastian Lopienski
Collaborator liaison(s)
Cris Pedregal, Alexandre Reigada, David Ebert, Vincent Leocorbo, Dmitrij Dolgušin

Collaborators

Project background

Today, high availability is a key requirement for most platforms or services. The Database Services group maintains critical services for CERN that are used daily by the majority of users. This project enables us to review the current system, assess its scalability and explore further capabilities by integrating them with new technologies.

Recent progress

For the integration exercise, the group is using the Oracle Cloud Infrastructure (OCI).

The project team successfully replicated its process for creating virtual machines on OCI. This involved registering public-cloud virtual machines on the CERN main network, as well as integrating them with CERN’s central configuration management system.

The ability to deploy machines on OCI within the CERN network enabled us to use OCI running Oracle databases like on-premises Oracle databases. Moreover, we automated the procedure for creating standby databases for our on-premises primary databases, and we configured a Data Guard broker for data synchronisation between them. We performed tests with different data sets to evaluate the performance on the network and storage sides.

The team maintaining Kubernetes applications investigated the container engine provided by Oracle for the Kubernetes (OKE) platform. We extended the application deployment process, which is now capable of deploying applications to the OCI running OKE platform.

Furthermore, using an Oracle REST Data Services (ORDS) application, we performed a complex integration test; for forwarding network traffic, we used a proxy server. The tests showed vulnerabilities and integration limitations for the applications. 

Next steps

We will continue to evaluate the performance of storage and network for databases and see what the best fits would be for our use cases. We will also investigate the cost of running the databases on OCI compared to running them in CERN’s data centre.

The members of the project would like to thank the support teams for Oracle Cloud Infrastructure and Oracle Terraform for their valuable assistance.