An Investigation into the Usage-trends of Canada’s Research Computing Clouds

Cloud platforms are gaining more footprint in many Information Technology sectors, including research computing. Infrastructure as a service (IaaS) remains a popular cloud delivery model for research computing clouds, enabling researchers to create isolated computing, storage, and network environments on demand. Despite the maturity of the IaaS model, cloud providers face several challenges due to the lack of insight into usage-trends of their platforms, including sub-optimal VM placement and arbitrary selection of adopted technologies. These problems become more difficult in the research computing domain, where the resource requirements and nature of computational workloads of VMs are highly diverse.In this paper, we illustrate the value of usage-trends’ analysis in VM placement decisions and technology adoption choices by IaaS providers. We examine the usage-trends of four of Canada’s research computing clouds, presenting our analysis of usage metrics that were collected since their inception and up to early 2019. We focus on examining the usage metrics with potential in guiding VM placement and technology adoption choices, namely the lifetime of VMs, the frequency of VMs’ creation, and the VMs’ resource allocations.