AWS for Industries

How to Build a Dynamic Supply Chain Platform: A Primer


To meet customers’ needs in a dynamic and fast-changing world, organizations must establish supply chain platforms that can proactively react in real time. Rapid technological development and advances in utilizing Internet of Things (IoT), artificial intelligence and machine learning (AI/ML), and cloud-based platforms can be leveraged to create scalable global platforms that capture data throughout the supply chain and intelligently react to events with minimal human intervention.

This post is the first in a series seeking to describe how AWS is helping clients solve supply chain challenges via these technologies. Here, we will focus on the fundamental building blocks required to build a dynamic supply chain, as illustrated in Figure 1, which minimizes disruption and waste and ensures customer SLA compliance. Subsequent posts will expand on more specific use cases and solutions.

Supply chain for goods transported by sea across two countries

Figure 1: Supply chain for goods transported by sea across two countries

A dynamic supply chain proactively reacts to events while providing end-to-end visibility and enabling communication and information exchange. The AWS cloud provides several services that enable this capability. These include collecting information from various supply chain systems and third-party sources in batch or real time, transforming the data to provide end-to-end visibility at every step, conducting analyses of necessary changes, sharing recommendations, and communicating changes back to the systems. Our services provide high scalability and availability with minimal latency while utilizing our global infrastructure spread across over 25 geographic regions. Tapping into this highly performant global infrastructure lets you reduce undifferentiated heavy lifting in order to focus on the core competencies required to manage the business. Figure 2 illustrates a sample reference architecture of a dynamic supply chain system incorporating some of these services.

Sample reference architecture for a dynamic supply chain

Figure 2: Sample reference architecture for a dynamic supply chain

The solution above builds on three essential foundations for a dynamic supply chain: event-driven architecture, real-time data collection, and end-to-end visibility and analysis. Below, we describe each one in detail, along with the challenges that they address.

Event-driven Architecture

A supply chain is a complex network with several integrated systems and thousands of data points. Considering the various systems in play, any process change becomes complex, requiring communication with and across multiple systems along the supply chain – drayage, ocean carriers, refrigeration facilities, equipment providers, and last mile providers, just to name a few. One challenge facing businesses is that these systems are often tightly integrated with each other. A change in just one system requires changes across multiple systems, resulting in latencies and delays. AWS provides numerous services, such as Amazon EventBridge, Amazon Simple Notification Service (Amazon SNS), and Amazon Simple Queue Service (SQS), that enable an event-driven architecture to decouple one system from another for increased efficiency and reduced risk of failure. An event is a signal that a system’s state has changed. Utilizing events for communication within a system is known as event-driven architecture. This reduces the need for tight integration across disparate systems, thereby reducing the risk of failure for large complex ecosystems, such as a global, complex supply chain.

In order to further reduce the risk of failure, event-driven architecture incorporates microservices to build the core functionality required in the system. The microservices approach breaks down large, complex code into small components that can be independently developed and scaled. Microservices can be deployed in AWS using serverless compute, such as AWS Lambda and AWS Fargate, or using server-based compute on Amazon Elastic Compute Cloud (Amazon EC2), Amazon Elastic Container Service (Amazon ECS), and Amazon Elastic Kubernetes Service (Amazon EKS).

Here is an example. When changes occur in the Warehouse Management system, such as a product stock level change, those changes trigger an event that is sent to Amazon EventBridge, which would then update data stored in Amazon DynamoDB. Amazon EventBridge can also begin a workflow that sends texts or emails for approval of additional supplies via Amazon SNS. Amazon SNS is a fully managed messaging service for both application-to-application (A2A) and application-to-person (A2P) communication. Amazon EventBridge would also pass on the event to a microservice that processes an order for that product by updating an API in a supplier’s procurement system. Your web or mobile application calls APIs in the procurement systems via Amazon API Gateway to order the product. Amazon API Gateway is a fully managed service that makes it easy for developers to create, publish, maintain, monitor, and secure APIs at any scale.

Real-time Data Collection

One of the most critical aspects of supply chain management is collecting data in real time from various points in the supply chain. This could be in a warehouse, port, or during transport from one provider to another from the point of origin to the final destination. IoT sensors attached to products or containers provide product data such as GPS location, temperature, air moisture levels, and light exposure. The sensors transmit this data in messages to AWS IoT Core, a service capable of supporting billions of devices and trillions of messages, and then routes those messages to AWS endpoints for storage and further processing.

In remote areas, reliable and sustained internet connection can be an issue. In these situations, utilize AWS IoT Greengrass to enable devices to act locally on the data they generate, thereby responding quickly to local events and reducing data transfer latency. For example, setup a device to reduce a refrigerator’s temperature if it increases beyond a certain point. Another frequent challenge is the lack of available cellular connections to provide the internet connectivity needed in remote areas. In that case, utilize AWS IoT Core for LoRaWAN, a fully managed feature that enables wireless sensors to transmit data via low-power, long-range wide area network (LoRaWAN) protocol.

Data must also be collected about external factors that could affect the supply chain. These factors may include a lack of transport availability due to weather conditions, crises such as COVID, or socio-economic events like strikes or government unrest. AWS provides several data collection services that specifically meet these challenges. AWS Data Exchange makes it easy to find, subscribe to, and utilize third-party data in the cloud. For example, use this service to subscribe to weather data from qualified data providers, such as Weather Trends International and CE Research Hub. Another example is a data set that lets you review the impact COVID-19 has had on businesses, along with economic indicator forecasts.

End-to-end Visibility and Analysis

A dynamic supply chain not only needs to provide timely and accurate information about the supply chain. It must also utilize analysis to uncover hidden trends and provide recommendations.

Users such as customer service operators, analysts, and executive management all need to be able to view updated information as well as review historical analysis like trends. In order to generate dashboards that provide this, you can load data from Amazon DynamoDB into Amazon Redshift and populate dashboards in Amazon QuickSight. Amazon Redshift is a cloud-based data warehouse that can query and combine exabytes of structured and semi-structured data using standard SQL. Amazon QuickSight creates and publishes interactive business intelligence dashboards and visualizations. Also, you can load data from Amazon DynamoDB into Amazon OpenSearch Service (successor to Amazon Elasticsearch Service) for users to search quickly for required information via the web application. Amazon Elasticsearch Service is a fully managed service that makes it easy for you to deploy, secure, and run Elasticsearch cost effectively at scale.

A dynamic supply chain enables utilization of Machine Learning (ML) for critical planning capabilities such as inventory forecasting. Traditional statistical models can handle forecasting for aggregated demand, basic trends, and repeated seasonal patterns, but they struggle when handling complex and inter-related data that may affect forecasts. For example, a statistical model may find it challenging to accurately predict inventory by accounting for dependencies between two products. This is where AWS services such as Amazon SageMaker and Amazon Forecast can be incorporated to predict inventory based on ML models. Amazon SageMaker provides purpose-built tools for developing complex and custom ML models for various planning use cases in supply chain predictive analytics, including inventory forecasting. Amazon Forecast is a purpose-built service that provides this out-of-the-box without needing to develop custom ML models. Choose between Amazon SageMaker and Amazon Forecast depending on the complexity of the use case, as well as your comfort level with ML. ML also enables Demand Sensing, the capability to anticipate and plan for short-term demand changes based on the current realities rather than solely the historical data. Demand sensing utilizes external data such as weather patterns, socioeconomics, geographic data, and social media to help predict and manage demand volatility.

Decision-making in a supply chain is a complex activity. It requires information and analysis, such as alternative ports, alternative transport mechanisms, availability of transport, time taken, and cost incurred. An ML branch called Reinforcement Learning (RL) provides prescriptive analytics for decision making by combining ML and Control Theory. Utilize Amazon SageMaker reinforcement learning, a service built on top of Amazon SageMaker for optimizing decisions such as route planning, in order to dramatically reduce the time taken to make decisions.

The Future of Supply Chain Platforms

To remain competitive in today’s complex global economic environment, businesses must increasingly develop dynamic supply chain platforms that react quickly to events in real time. This post has described how AWS Cloud services provide the foundations for developing a dynamic supply chain that minimizes human intervention, generates efficiencies, and reduces costs. In the next blog post, we will dive even deeper into specific problems and solutions, such as managing a supply chain to handle adverse weather events and managing a cold chain to ensure product quality.

To learn more about how supply chain management platforms are evolving to meet the demands of organizations in today’s dynamic global economy, stay tuned for the next installment in this blog series. And, if you’re ready for your own digital transformation, then AWS is here to help. Contact your account team today or visit our Retail webpage to get started.

Sanjeev Pulapaka

Sanjeev Pulapaka

Sanjeev Pulapaka is a Senior Solutions Architect in the US Fed Civilian SA team at AWS. He works closely with customers in building and architecting mission critical solutions. Sanjeev has extensive experience in leading, architecting, and implementing high-impact technology solutions that address diverse business needs in multiple sectors, including commercial, federal, state, and local governments. He has an undergraduate degree in engineering from the Indian Institute of Technology and an MBA from the University of Notre Dame.

Dnyanesh Patkar

Dnyanesh Patkar

Dnyanesh Patkar is a Senior Practice Manager, Supply Chain and Logistics at AWS. He works with customers to envision, develop, and execute on transformational business and operating model strategies in this space. He is a seasoned executive, and his background includes leading business transformation programs, serving in general management roles with P&L responsibility, and creating high performance teams. He has over 25 years of experience at companies such as Schneider National, DiamondCluster, and National Semiconductor. Dnyanesh received his MBA from the Wharton School of Business and a M. Engineering from Cornell University.