Cloud computing has been around for some time and is firmly embedded in our everyday digital life. For those building cloud applications, there are roughly two main approaches:
Cloud-native: Sometimes classified as ‘platform-as-a-service’ or ‘serverless’ capabilities, this first approach takes advantage of services intrinsic to the cloud. Typically, these services exploit the cloud’s innate architecture that separates compute from storage to create better availability, affordability, improvement velocity, and scaling that isn’t possible with on-premises packaged applications and raw hardware.
Cloud-hosted: This approach helps users quickly deploy on-premises, packaged software directly in the cloud using rented host machines to speed up operations.
Built using cloud-native architecture, Moody’s RMS Intelligent Risk Platform™ (IRP) takes special advantage of some of the cloud’s powerful capabilities.
In this post, we will look at why Moody’s RMS chose to rebuild our analytics in the cloud using a cloud-native approach as described previously and to outline the advantages this approach gives to our customers.
First, why did Moody’s RMS take the long route? Why not just rent machines in the cloud, quickly wrap a new skin on our existing software applications like Moody’s RMS RiskLink® or Risk Browser™, and simply call that our new cloud platform?
There are some big downsides to a cloud-hosted approach, as it would simply subject our customers to the same limitations on-premises technologies have today.
Let’s take a deeper look at the advantages you get with the Intelligent Risk Platform and its cloud-native architecture – and what’s different compared to a cloud-hosted approach.
Figure 1: Cloud-hosted versus cloud-native architecture
Better System Availability
Even with a cloud-hosted approach, the on-premises systems that they are based on and use will still need regular maintenance, therefore downtime periods must be factored in. Being cloud-native, the Intelligent Risk Platform does not require those dreaded hardware or software maintenance tasks.
No hardware replacement, no operation system patching, or hotfixes, IRP applications also come with service level agreements (SLA) that promise uptime guarantees.
And when a platform application is updated, with many of these platform apps receiving updates every four to six weeks, customers do not experience any downtime during these updates.
Better Affordability
An important consideration for any cloud-based solution is cloud costs which are typically calculated by consumption in seconds. A twenty-second utilization of a service typically costs twice as much as a ten-second utilization of the same service.
This creates a large disadvantage for the cloud-hosted approach, as similar to using on-premises hardware resources, users must keep most of their infrastructure online all the time – and pay for it.
In the graph below (figure 2), the red lines represent the overcapacity being paid for in a single day when hosted machines are up and running all the time.
Zooming out from a single day to a full year, and considering busy renewal periods, the area under red adds up to an even larger cost burden which gets translated into the vendor’s solutions pricing.
Cloud-native approaches are represented by the blue areas in the diagram and ensure costs for compute are not incurred when a system is quiet.
When a system isn’t active running jobs such as model execution, geocoding, hazard lookup, accumulation, marginal impact, import, export, data conversion, accumulation, exposure or loss roll-up, and many more, we do not pass on costs for just keeping machines running.
Figure 2: Cloud-hosted capacity usage versus cloud-native capacity usage
Better Scaling and Throughput
The separation of compute and storage in the cloud creates a new architecture that enables new types of scalable systems. Thanks to the scale available today, we can build modern, complex analytics systems, which are affordable.
These systems help open new insights, enable models to more realistically represent losses, run analytics on an entire book of business for the largest (re)insurers and brokers in the world, and support more users without slower response times under heavy loads.
Side-by-side Model Versioning to Facilitate Change Management
With frequent model and application updates, our customers enjoy seamless updates without worrying about the impact on their existing work. This is all possible thanks to the scale that comes from a cloud-native approach.
Whether it is a new model version, enhanced data set and a new dashboard, or a report in our apps, new updates simply show up within dropdowns on the user interface and in the APIs. And thanks to side-by-side versioning, there are no changes to financial results between these updates.
For example, each year Moody’s RMS delivers the latest version of RiskLink. In June 2023, version 23 will be delivered to our on-premises clients for installation, but for our IRP clients, version 23 simply shows up on the dropdowns within our Moody's RMS platform applications such as Risk Modeler™, TreatyIQ™, UnderwriteIQ™, ExposureIQ™, Location Intelligence, and so on.
The latest version 23 appears right next to all the other RiskLink versions from version 18 to version 22, and customers get to choose when to upgrade by selecting the new version themselves.
Connected Applications Increase Consistency in How You Analyze Risk
Directly related to the above point, without a cloud-native approach, it would not be possible to build multiple applications with shared services.
Figure 3: Moody's RMS Intelligent Risk Platform
In the platform architecture shown in the diagram above, you can see the platform applications at the top row of the diagram; applications that are built upon the IRP’s range of microservices.
The set of shared services in the red and blue boxes in the middle of the diagram provides a consistent set of services to these apps. For instance, there is a single geocoding service delivering a consistent resolution of locations regardless of the app you use.
A robust financial model processes policy terms, treaty terms, and marginal impact, and a consistent set of data management services enables the importing and exporting of data in various well-known formats as well as for many RMS models that run under various model frameworks.
This helps ensure that there is a consistent approach for how you are managing risk across its lifecycle – from underwriting to portfolio steering and risk transfer.
Not only are microservices built to be shared, but data in the platform is also shared as well. In fact, all these applications ‘speak’ our Moody’s RMS Risk Data Open Standard (RDOS) when they interact with data within our platform.
This makes it possible to import Exposure Data Modules (EDMs) from one app such as Risk Modeler and then also have that EDM available in ExposureIQ, and so on. This unified risk analytics architecture would not be possible if we had not built our entire architecture using the cloud-native approach.
What About Backward Compatibility?
I do realize that despite all the advantages of the cloud-native approach, some customers may find the change to the cloud to be hard. That’s why we have built bridging solutions like Moody’s RMS Data Bridge that takes your existing RiskLink instance and data into the IRP, to preserve your on-premises investments while migrating to the cloud.
Changes to results and outputs can also be hard to manage when moving to a new solution. That’s why with platform applications like Risk Modeler and others, the IRP guarantees matching results with our on-premises solutions.
Conclusion
Certainly, there are shortcuts that can be taken to present some old on-premises applications as new cloud applications by simply renting machines in the cloud.
However, it does not address the fact that you will have all the same current limitations as you have on-premises. In fact, you will find your availability, your costs, and your analytical throughput will suffer greatly with these architectures just like on-premises deployments.
More importantly, you will have applications that just silo their data and can’t connect effectively with other applications, which could mean you will have to build all the data connectivity yourself.
With the IRP, Moody’s RMS did not take those shortcuts because we believe the advantages of the cloud-native approach are too ‘expensive’ to miss.
To help you with your decision on whether a cloud-native or cloud-enabled approach is right for you, we built the Cloud Risk Analytics Buyer’s Guide, which can help you first ask the tough questions required so that you can realize the full value of your data, generate new insights, and take advantage of what the cloud has to offer.
Managing Director - Head of Product for Moody's RMS
Cihan Biyikoglu is the Managing Director - Head of Product for Moody's RMS, responsible for product management across the full suite of Moody's RMS models and risk management tools. He has extensive experience in leading product management for innovative machine learning and big data analytics solutions at Fortune 500 companies over the last 20 years.
As a former Vice President of Product at Databricks and Redis Labs, Cihan developed the product strategy and road map for open-source technologies such as Apache Spark and Redis and respective enterprise offerings in the public and private cloud platforms.
Cihan also worked on products at Microsoft, Couchbase, and Twitter, where he focused on on-premises and cloud offerings in the data and analytics space. At Microsoft, Cihan focused on the incubation of the Azure Cloud Platform in its early days and the SQL Server product line, both of which have grown into multi-billion-dollar businesses for Microsoft.
Cihan holds several patents in the data management and analytics space, and he has a master’s degree in database systems and a bachelor’s degree in computer engineering.
