RMS Selects OCF for Double Petabyte Storage Infrastructure
Monday 31 October 2011
RMS Selects OCF for Double Petabyte Storage Infrastructure, Leveraging DDN Technology for Scalability and Performance
Risk Management Solutions (RMS), a provider of products, services, and expertise for the quantification and management of catastrophe risk, invests in a new double petabyte storage system. This enables RMS to meet the massive and increasing storage, rapid recall and data archiving needs of its Global Catastrophe models, which helps insurers and re-insurers calculate risk on their portfolios.
The pioneering storage system was designed, installed and configured by high performance data management, data processing and storage provider OCF www.ocf.co.uk. The storage environment replaces an open source system, pools storage hardware and reduces data backup time by 50 percent.
Storm risk in Northern Europe is one of the most complex climatic phenomena in the world to model. RMS has developed its European Windstorm Model, The RMS Windstorm Model, which uses a blend of numerical and parametric modeling, to create stochastic windstorm tracks. Whenever a large storm strikes, insurers run a risk assessment based on simulation runs, pre-computed and plot all their assets and points of interest into the model and calculate their exposure in the affected area. The European Windstorm Model enables insurers and reinsurers to understand and quantify what the risk would be to their portfolio should an event occur.
Ben Morfett, IT Manager comments: “Based on the complexity and granularity of our next generation catastrophe models, which has a direct consequence on our storage capacity and back up and retention plans, we required a solution that would enable us to better respond to the changing needs of our customers.”
“We have seen growth in raw storage increase 10 fold in just two years, and this required a step change in how we address our data management needs. Working with OCF, we have implemented an enterprise storage solution that maintains the consistency and integrity of our data backup and recovery point objectives - ensuring we could sustain operations by recovering from a potential disaster ourselves.”
RMS continues to invest in its compute infrastructure and has over 800 cores. IBM’s General Parallel File System now sits across all servers, each of which has direct access to the new central storage pool across Infiniband and has an aggregate peak performance of 2.76 Gbps and an individual host throughput of 1.3Gbps. GPFS and IBM Tivoli Storage Manager (TSM) work side by side and all data is backed up to a 900 slot SpectraLogic T950 tape library with a total storage capacity of 2 Petabytes.
Ben comments: “With TSM’s ‘Incremental Forever’ we only really need to run one single full data backup, which takes about 3.5 days to complete, down from 7 days. As GPFS knows which files have changed, it instructs TSM to only backup those altered files – as a result our nightly backup has reduced from 15 hours to 3.5 hours. Ultimately this means simplified management and ensures far greater data integrity.
By using TSM, the tape library can be used as an additional offline storage pool, enabling RMS to expand its disk storage onto tape and therefore reducing archive and recall time. Additionally, any data that is not required online can be seamlessly moved to tape thus freeing up valuable high performance disk space.
Ben concludes: “OCF has done a great job and their work here is serving as the blue print for our sister data centre in California. The GPFS and TSM solution is now part of a much wider multi-site data replication plan. Therefore, in the event of any one of our datacenters going offline, the integrity of RMS data can be assured.”
Commenting on the new architecture, Julian Fielden, Managing Director at OCF, says: “One of the greatest challenges for any industry is its silo approach to unstructured data. RMS has the perfect architecture that can be expanded easily as the business grows.”