The trading arm of a large UK investment bank needed to expand its high performance computing (HPC) grid to absorb the trading levels of its newly merged organization. Due to space constraints and an objective to reduce costs, the bank wanted to outsource the grid from its City of London location to a new, more cost-effective datacenter outside the M25. The primary goal of the move was to find the right balance between performance and cost, while ensuring that this highly sensitive trading environment was not affected by any latency issues. Because of the major business risk this presented, the IT team needed to provide assurances that end-user performance and links to external trading systems were not going to be impacted.
The bank believed the use of Sumerian’s CPaaS would allow them to accurately scenario model all aspects of the planned migration and successfully de-risk this major £300M investment program.
Sumerian created a Baseline model of the existing trading environment grid infrastructure and the applications under migration consideration. The model included detailed measurements of existing application performance and task execution time to establish a baseline of pre-change application latency and throughput.
Using this model Sumerian then applied predictive scenario modeling to determine whether application performance would be impacted by any latency issues from a move to the new datacenter location, and what type of datacenter hosting configuration was required to achieve the most optimal balance between performance and cost. For example, whether applications should be hosted solely at the new, more cost-effective location, spread across both locations or remain at the existing datacenter. From the results of this analysis Sumerian was able to recommend the most optimal post-move deployment architecture, including which parts of the infrastructure and component software needed to be altered to meet latency and throughput requirements in the post-change environment.
Outcome and results
Sumerian’s CPaaS showed that latency impact varied across the application suite, and also depended heavily on where the latency was introduced. As a result, Sumerian was able to make specific recommendations on how to rightsize capacity at the new datacenter and also how to deploy the applications in a way that avoided any impact to the business thereby de-risking the migration.
In parallel, Sumerian was able to make specific recommendations to the application development teams on the amount of capacity required to ensure all applications were optimized to reduce their exposure to latency. These recommendations were then implemented by the development teams in advance of the move.
Above all, by using Sumerian’s CPaaS to quantify and model the applications’ latency requirements against the various scenarios under consideration, the IT team gained the validation it needed to ensure the new outsourced grid datacenter could uphold levels of performance demanded by the business – but also realize multi-million pound cost savings.