AWS for Industries

Modernize your retail edge architecture with microservices

Technology in most retail edge locations (i.e., stores, distribution centers, factories) is a mix of disparate applications and infrastructure, creating a variable IT landscape. This web of complexity prevents companies from quickly and consistently deploying new systems across edge locations. That begs the question: How can retailers manage edge infrastructure at scale with consistency and reliability? The answer: cloud-based microservices integrated with edge infrastructure.

The following five strategies can help your company implement an agile microservices architecture that allows you to deliver consistent experiences to reinforce your brand and exceed customer expectations regardless of the sales channel.

  1. Rethink your store application architecture

Traditional application architectures create silos that fragment the brand experience for shoppers. Store applications, mobile, and online apps are typically developed independently. This creates complex extract, transform, load (ETL) challenges to manage customer data and prevents retailers from gaining a holistic view of customers across channels.

With a microservices architecture, you can integrate all of your retail sales platforms—in-store, mobile, and online—with a set of centralized business processes. If you aren’t familiar with microservices, think of segmenting distinct processes into reusable components, like a shopping cart, pricing, promotions, and payment transactions. As you deploy standardized microservices across platforms, functionality is consistent from channel to channel. When you release a new feature, an additional payment option for example, it’s available across all channels. On top of that, you can simplify application management and reduce IT costs.

A microservices strategy can lay the foundation for consistent, cost-efficient shopping experiences, and products like Amazon API Gateway, to integrate existing systems, and AWS AppSync, for real-time bidirectional synchronization across channel applications, should be part of your deployment plan.

  1. Enable remote IT management with microservices and tools

As I mentioned, microservices are the foundation for consistent customer experiences. They are changing the traditional approach to systems management in stores too. Remember when I said in the section above that retail systems are usually siloed? This bespoke lack of integration makes it difficult to manage applications and devices. Remote management tools and processes are generally unique to each application, or worse, IT needs to visit stores for installs, upgrades, and troubleshooting, which can be very costly.

A microservices architecture changes the game, enabling you to simplify IT management—so you can remotely deploy and manage applications and hardware. That means you can use microservices and a consistent set of tools to standardize deployment processes across all retail stores, giving you a streamlined, cost-effective alternative to traditional bespoke IT architecture.

With tools like AWS Snowcone, AWS Systems Manager Agent, and AWS IoT Greengrass, you can automate and consistently manage installs, updates, and configurations from a central cloud-based control plane.

  1. Deploy cloud-based IoT and vision sensors in edge locations

It’s hard to track customer behaviors in brick-and-mortar retail locations. That’s where real-time IoT and vision sensors can help retailers understand what people are doing in stores, with what products people are interacting, and where shoppers spend time. This kind of powerful analytical insight can drive product placement, in-store advertising, queue management, and stock replenishment.

In the past, retailers have deployed standalone IoT sensors or vision systems in stores. The problem is that each system is a silo that has to be managed separately. Also, the different systems have their own machine learning (ML) infrastructures and algorithms, which don’t give you a holistic picture of customer behavior across all stores.

Cloud-native IoT and vision systems have changed the game. These systems enable you to better manage and optimize bandwidth requirements for edge locations. To mitigate bandwidth challenges, these applications should handle inferencing locally, meaning events are captured and analyzed with ML models in stores, batched, and sent to the database in smaller payloads. A tool like AWS IoT Greengrass does exactly that. Since only analytical results are transmitted over the network, you can reduce bandwidth costs. A microservice in the cloud can aggregate and analyze IoT and vision data from all edge locations to help you make data-driven decisions. And in addition to events flowing to the cloud from the edge, bidirectional communication microservices can push events to edge locations in real time.

Diagram: Retail Computer Vision Edge Inference Reference Architecture

This reference architecture describes how retailers can use computer vision models deployed on edge devices to capture insights at the edge. Insights are sent back to the cloud to drive real-time business processes and power long-term analytics.

  1. Use microservices to manage application data during internet disruptions

Retail locations often struggle with unreliable internet connections—not ideal for applications intended to operate online. To ensure reliable functionality, applications can be designed to run online and offline.

In the past, retailers designed individual applications to collect activity locally/in cache during network outages. Although this approach works, it makes your software footprint more complex as each application is managed separately. A more efficient method uses microservices to store downtime application data at all retail stores.

With AWS Stream Manager, you can deploy edge applications to eliminate this durability concern—to batch transactions, messages, and data feeds when there’s a network failure and automatically forward information when connectivity is restored. This means you can continue to accept payments, capture video streams, and collect inventory data even when retail locations lose internet connectivity.

  1. Centralize microservices technology procurement and vendor management

With a microservices architecture, you can easily procure solutions developed by vetted third-party vendors, such as AI/ML models and containerized applications, to run at the edge. Because the microservices are standardized, your IT team can deploy them in the cloud and at edge locations much more quickly, which increases efficiency and productivity.

AWS offers an AWS Marketplace of ISVs that publish software for AWS managed edge devices in stores, distribution centers, and factories. On top of that, when you centralize technology procurement in the cloud with AWS, you can maximize your buying power, purchase at scale across all edge locations, and take advantage of negotiated bulk discounts.

Build your edge technology foundation with microservices

Many retailers are replacing legacy technologies with applications designed with microservices—to drive consistent shopper experiences, reinforce their brands, streamline IT application management, and reduce operating costs. By using a cloud-connected edge strategy with microservices, store managers can focus on customers, merchandising, productivity, and profitability, instead of struggling with technology.

Ask how AWS can help you make it happen and plan to attend AWS re:Invent 2020, a free virtual conference, to learn how to innovate and transform your business with the power of cloud.

Paul Fryer

Paul Fryer

With an illustrious track record as a technology thought-leader, Paul Fryer joined AWS in 2016 and is the worldwide technology leader for retail. He’s responsible for defining and executing the company’s retail technology strategy, which includes building retail-focused solutions with services like Amazon Forecast and Amazon Personalize, as well as experiences like Frictionless Shopping with AI/ML and IoT services from AWS. Paul also works very closely with AWS global retail customers to help transform their businesses with cutting-edge AWS technologies. Prior to AWS, Paul worked at Fiserv and was responsible for banking software architecture with global financial institutions.