Moody’s Analytics simplifies and streamlines risk modeling with Azure HPC

Critical computing capacity

As insurance industry regulatory requirements expanded and the financial reporting requirements for compliance increased, Moody’s Analytics recognized that providing a pathway to support an exponential increase in demand for compute capacity would be critical for its customers. “Actuarial systems need a very high level of compute for calculations that are required to meet industry regulations  such as the International Financial Reporting Standards,” says Victor Rubinstein, Senior Director of Technology at Moody’s Analytics. “We had to scale our distributed processing engine to perform these calculations in the elastic cloud environment, as in many cases they can no longer be completed on-premises."

Moody’s Analytics grid computing for risk modeling journey started with distributed calculations in on-premises server farms that often needed to be overprovisioned to meet peak demands and refreshed on a regular basis. To optimize the task distribution process, Moody’s Analytics developed AXIS GridLink to manage the workflows used to process calculations and create efficient and highly secure production environments. As the industry moved from physical hardware to virtual machines (VMs), AXIS GridLink developed the ability to control both on-premises and cloud clusters. Moody’s Analytics then adapted its AXIS GridLink software to run on thousands of cores in Microsoft Azure, creating a risk simulation compute platform called GridLink-as-a-Service (GlaaS) that’s underpinned by Azure Virtual Machines and Azure Virtual Machine Scale Sets.

One-click risk modeling

Designed on a pay-per-use model, GlaaS serves as a cost-effective option for Moody’s Analytics customers to take full advantage of the latest cloud networking capabilities, virtual server instances, and storage technology for actuarial modeling jobs. “With the click of a button, GlaaS uploads data to Azure, provisions the necessary cores, deploys the software, runs the model, and packages the results for end-user consumption,” explains Rubinstein. “We’ve onboarded many clients to GlaaS, and we’re successfully running a massive amount of core hours on Azure.”

By using Azure Virtual Machines, GlaaS can now automatically spin up thousands of processor cores within minutes to do batch calculations for clients, providing results within a few hours versus sometimes month-long run times using on-premises compute resources, along with the ability to automatically scale down capacity when it’s not needed. With the scalability of Virtual Machine Scale Sets, customers can provision significantly more compute capacity on demand and broaden the scale and speed of those calculations, improving accuracy and profitability. That elastic scalability helps Moody’s Analytics meet its customers’ needs quickly, accurately, and cost-effectively.

For each requested job in GlaaS, the service automatically manages the preparation of dedicated grid capacity in the cloud and performs the calculations, requiring no intervention from customers’ technology teams. Because there’s no competition for disk bandwidth or computing capacity, customers can better allocate cores and respond to surges in demand, meet reporting deadlines, and manage costs. “Thanks to the extensive capabilities and ease of use of Azure, GlaaS removes all the complexity for our customers,” says Rubinstein.

Meeting customers where they are

Because GlaaS customers have varying needs for computing parameters, storage, and the number of VMs required for any given job, Moody’s Analytics ensures that each customer’s setup is individually configured. It uses an algorithm to automatically predict the required disk size for each job based on the size of the uploaded data. Customers with high demand for storage capacity and performance benefit from Azure Disk Storage added to the scale set VMs. GlaaS also takes advantage of the deployments in multiple Azure regions to meet the data residency requirements of Moody’s Analytics customers and also provide best performance when it comes to data transfers.

GlaaS runs on high-performance SKUs, including Azure HC44rs VMs optimized for high-performance computing and compute-intensive Fsv2-series VMs. Customers that might have had access to 1,000 cores on-premises can go big with Azure, onboarding up to 10,000 or 20,000 cores, running jobs in parallel on different dynamically provisioned grid farms, and creating enormous performance gains. To make performance fast and economical, Moody’s Analytics is continually monitoring availability of new, faster, and cheaper SKUs that may improve performance and reduce cost for customers. The company also taps into Log Analytics in Azure Monitor to track trends in VM performance and respond accordingly.

Although many Moody’s Analytics customers have fully moved to GlaaS in the cloud, some are continuing to take advantage of on-premises infrastructure investments. Thanks to the company’s promise to meet every client where they are in their cloud journeys, these customers are still able to benefit from the system’s newfound elasticity. “Users who still have on-premises grid farms can use GlaaS as supplemental computing specifically for those bursts at the end of the month, end of the quarter, and end of the year,” says Sean Welsh, Director,  Insurance Modeling and Analytics at Moody’s Analytics. “Ordinarily, they would have had to make an enormous capital outlay to do this, if it were even possible, but we sort that all out for them.”

Making more with less

It was important to Moody’s Analytics that its customers be able to reap the solution’s benefits without having to take extra steps to meet their objectives. With Azure, the company has the flexibility to support multiple platform configurations and offer GlaaS fit to purpose for any customer. “At Moody’s Analytics, client satisfaction is number one,” says Rubinstein. “With Azure, we’ve been able to provide them with the critical compute they need in an enormously quick and simple manner, and we can provide continuous service to the solution.”

Thanks to its innovative Azure implementation, Moody’s Analytics is driving growth, enhancing operating efficiency, and delivering value faster than ever before, not just for its customers but in-house too. In addition to ensuring customer satisfaction, selecting Azure for GlaaS has helped the company’s IT team make more happen with remarkably less work needed to reach new milestones. “Everything with Azure is more unified and efficient, and if there’s any issue, we can deploy one fix for the entire region without affecting everybody else,” says Rubinstein.

Find out more about Moody’s Analytics on Twitter and LinkedIn.