Docker-Based Automatic Deployment for Nuclear Fusion Experimental Data Archive Cluster
Qiang Liu, Wei Zheng, Ming Zhang, Yuxing Wang, Kexun Yu · IEEE Transactions on Plasma Science · 2018
With the rapid growing of experiment data, using of distributed database or file system is the trend in fusion storage. For such storage systems, a cluster is required, but its setting up and services deploying are challenging. Installing and configuring the operating system and software runtime environment and managing all the dependencies are a complicated task. Apart from that, due to the variety of server hardware and difference between test and production environment, particular setup procedures are required. J-TEXT cloud database (JCDB) is a nuclear fusion experimental data storage and management system, aiming to satisfy the requirements of future long-pulse experiment, powered by MongoDB, Cassandra cluster, and Web applications. It is not an easy job to setup JCDB on these servers from scratch or to scale out the system. The system uses Docker, a lightweight software containerization platform, to build cloud database and Web applications, guaranteeing consistency of the all servers and automating the deployment and scale-out tasks, reducing plenty of time. JCDB pushed two images to Docker repository and scaled out by pulled specified images. Even without any physical servers on hand, you can also deploy it on cloud services such as Amazon Web Services and Microsoft Azure.