Yuji Ikeda

Since it was necessary for laboratory staff members to place an order and execute settings in each department to use the cloud service, it has been necessary for them to spare time in developing the environments. In addition, if different equipment is installed depending on a case, when an unexpected situation such as failure occurred, it also took time to restore it. Under those circumstances, if we create a specialized team for infrastructure development to support research activities in the laboratory aiming at integration, we believed that quick operation at lower costs would be possible. That is why we launched the “XFARM” service.

Our work begins with setting the service ready for use based on a service order issued by each department. There are many kinds of requests, such as using as many servers as possible, preparing network environments that enable even remote operation, and using a server with GPU installed. In “XFARM”, we proceed with the introduction from purchasing necessary equipment to building the system environment, taking account of combinations with external services. Bulk purchasing of required equipment for our laboratory was also effective in reduction of costs. Since any department needs not to have less frequently-used equipment, collective ordering has enabled avoiding duplicated ownership. In addition, since we can collectively manage information or operation manuals, etc. for the installation, we can also quickly respond to failure.