Why Highly Available API Architecture Is Vital for Global Organizations

For more than 180 years, Dun & Bradstreet has supplied crucial data that helps our customers make informed decisions. Originally, this data came in the form of paper reports, and much more recently, it is delivered via APIs. As the means of delivery has advanced, so too have customer expectations. These days our customers are global, and whether they're in Dallas, Dublin, or Delhi, they demand quick response times and improved stability.

The D&B Direct+ suite of APIs was created to meet these needs and has continued to evolve to improve the customer experience. We achieve this through a relentless focus on low latency and high availability.

What Is D&B Direct+?

APIs (application programming interfaces) enable software applications to communicate and exchange data in a standardized manner, and D&B Direct+ is a suite of APIs providing multiple benefits that include:

  • Global perspective – Our APIs access the Dun & Bradstreet Data Cloud, which is the largest set of B2B data available, covering hundreds of millions of businesses worldwide.

  • Data freshness – We continuously monitor for changes that impact information in the Dun & Bradstreet Data Cloud, verify such changes, and update data accordingly.

  • Scalability – D&B Direct+ supports almost one billion requests per month, and the capacity of our APIs adjusts based on usage.

  • Availability – Our objective is to provide customers with an “always on” service, also known as high availability.

Low Latency

In general, the greater the distance is between a customer’s physical location and a server, the longer the response time will be. To improve response times for our global customers, the D&B Direct+ Data API is deployed in four regions around the world. We use latency-based routing to determine the best region to serve the request, so customer requests automatically route to the nearest data center for a speedy response.

In addition to more localized support for customers, Dun & Bradstreet’s strategy to minimize processing time includes specially designed architecture. By regularly precomputing much of our data, we can reliably achieve processing times well within 100 milliseconds. At present, our processing times average just 58 milliseconds.

The combination of multi-region deployments and low processing times results in low response times. This helps our customers leverage our APIs for a variety of use cases, including those that require response times measured in milliseconds. The graph below provides more detail on our average response times; note that response times will vary based on a customer’s location and the quantity of data returned.

High Availability

High availability enables a system to continue to respond to requests despite the failure of one or more components. This could include the failure of a random server, a database, a data center, or in extreme cases, an entire geographic region. In more specific terms, high availability means:

  • Eliminating single points of failure within your architecture

  • Designing your system to scale out based on demand

  • Automatically recovering from failures (without human interaction)

  • Utilizing a load balancer to direct traffic to healthy servers (by deploying in multiple independent data centers)

D&B Direct+ APIs achieve this fault tolerance by using a multi-region, active-active architecture that helps ensure all of the above. D&B Direct+ APIs are cloud-based, using Amazon Web Services (AWS). In AWS parlance, they use:

  • Auto-scaling groups to replace unhealthy instances and scale compute capacity

  • Multiple availability zones to guard against the loss of a data center

  • Multiple (active-active) regions to help protect us in the unlikely event of a region failure

  • Route 53 latency-based routing to automatically direct user requests to the best region

How Dun & Bradstreet Routes and Balances Requests

Several D&B Direct+ APIs use DynamoDB, which is automatically replicated across three availability zones. Additionally, this database is deployed in four AWS regions providing a total of 12 copies of the data. In the unlikely event of catastrophic loss or corruption of all 12 copies, Dun & Bradstreet retains the ability to rebuild the database. This extreme level of redundancy is fundamental to our business continuity strategy and a key differentiator in the marketplace.

Just as the mechanism of delivering data has advanced and accelerated during the past 180 years, Dun & Bradstreet will continue to progress D&B Direct+ to help ensure ever-increasing availability to our customers. Will every application need this level of availability? As in most things, it depends on a cost/benefit analysis and the specific use-case. However, experience has taught us that it's worth the investment many times over.

Want to know more about D&B Direct+? Visit our resources for details and to connect with our data and analytics experts.