The Infrastructure Group (IG) have been working over the last twelve months to bring Openstack Cloud Software into production internally within TSSG.
Deploying Openstack will not only bring the benefit of reduced infrastructure costs but there we should also witness a noticeable improvement on productivity within TSSG research units. The current cloud system within TSSG does not offer self-service and instead delivers a time-consuming process between developers and the infrastructure group.
What has been noticed over the years is that the lack of self management for the developer results in an increased life-cycle of virtual machines and backups, which in turn increases management complexity and wastes valuable resources.
The Openstack cloud aims to resolve this by allowing researchers within TSSG to manage their own services which includes; virtual machine creation/termination, backups, ip address allocation, ram & cpu allocation, whilst allowing the user to monitor and stay within their given resource quota.
Once the user is familiar with the Openstack interface, services can be created almost instantaneously and many users are already integrating tools such as; Docker and Ansible Playbooks to setup their own services when, for example a new project starts up.
The desired outcome is to create a standard infrastructure process for each unit within TSSG. At the start of a new project, standard templates, images and containers can be used to quickly setup all that is required e.g: Web Server, Redmine, Gitlab, Build Server and project VMs allowing the developers to focus on the the project tasks.
Bringing Openstack into production has been no trivial task, a number of systems have been released to a subset of TSSG users over the last twelve months that has allowed close monitoring of the system performance and has helped identify weaknesses in the system. The latest Openstack system is under the testing phase at the moment will be ready for all TSSG staff within the next few weeks. A later post will give a technical overview of how the final install has dramatically improved network performance, redundancy, bottlenecks and SPOFs.