HPC and the cloud
Capital markets participants who already process large volumes of data will most likely already be utilising the power of on-premise grid computing to undertake large computational calculations, although the processing capability is limited by the physical compute size. Most on-premise grids are sized to be able to deal with standard capacity, plus an overnight spike in processing for simulations and reports, but cannot easily scale to accommodate potential reruns arising from `data errors’, or to cope with new regulatory calculation requirements.
Such challenges arise from the use of an on-premise technology stack, where there is always the risk of excess capacity at most times and insufficient capacity at others, with the danger of new completing processing by the commencement of the next trading day. We believe with the right cloud strategy these risks can be eliminated. How?
By deploying grid processing nodes to the cloud, capital market participants are able to ‘spin up’ enough compute capacity to deal with business processing during the day, then augment this overnight or when additional large and complex calculations are required for compliance purposes. HPC is the perfect use case for utilising the elasticity of the cloud and there is little doubt that migrating HPC to the cloud is essential to simplify regulatory compliance, mitigate the risk of non-compliance and ensure compliance is embedded within business strategy.