Doubling daily e-commerce through OMS modernization with K8s
Challenge
The Infosys client – a large distributor of heavy-duty trucks and trailer parts and services in the USA – is focused on driving revenue growth through digital demand in the business-to-business and business-to-consumer space.The program objective was to enable optimized Omni Channel orchestration and fulfillment across all channels.
The client faced a number of challenges including lack of Omni Channel capabilities; inability to fully utilize network inventory; complex requirements of automotive parts distribution; complex Integrations across legacy and modern technologies; and building an integrated solution across multiple products and technologies such as order management system (OMS) in Microsoft Azure Cloud (using Kubernetes-compliant RedHat OpenShift platform) and ecommerce websites. Disconnecting the outdated system and offering a fresh approach to ensure efficient operations also posed a challenge. There was also a lack of a modern OMS and no sourcing optimization on the legacy ERP system, making it difficult to deliver a best-in-class online shopping experience. It was vital to be able to move to a new platform which retained functionality without any loss of data across systems, all while the platform retained 24×7 availability. And, finally, there were complex manufacturing requirements, including challenges with inventory visibility because there was no single way to cross reference parts across various enterprises.
Given these hurdles, it was imperative for the client to embark on OMS modernization using containerization to enable an integrated omni-platform that would be reusable across channels.
Solution
Considering the complexity of the tech-stack, the product team worked with our solution implementation team to onboard customer on the latest version of the OMS deployed on MS Azure with industry leading Red Hat OpenShift Container Platform (RHOS). No other team had implemented this kind of solution and it was in a true sense the first-of-its-kind solution. The solution is available as a multi region/multi availability zone Azure Cloud offering. The team engaged with multiple groups to implement the OMS solution and rolled-out it to B2B and B2C business lines integrating with a number of trading partners and multitudes of drop ship vendors.
Impact
The client:
- Doubled eCommerce daily sales through the single view of inventory across channels
- Implementing OMS as a containerized application (being a step closer to carbon neutrality) on Kubernetes compliant ARO platform reduced dependency on static infrastructure and effectively utilized resources, minimized waste, and, through energy efficient data centers, reduced IT carbon footprint.
- Increased ROI of digital spend by 5x through seamless order orchestration integrating modern applications to legacy ERP stitching supply chain components together
- A single solution catering to multiple business lines resulted in a 75% reduction in external partners’ integration with digital ecosystem;
- Better fulfillment optimization leveraging inter-store networks to consolidate shipments reduced carbon emissions
- A 20X increase in SKUs exposed through inventory optimization;
- Containerized ARO deployment with automated CI/CD pipeline enabled a 60% reduced deployment cycle times and additionally resulted in a huge reduction in person hours (and thus annual savings)
- Accurate fulfillment options with 95 percent of calls responding under 5 seconds
- The solution provides a ready-to-implement template-based approach enabling at least a 25% reduction in timeline and reduced carbon emissions for project execution.
- Automated complex 200+ inter-store pick up networks resulted in optimized customer shipments and reduced carbon emissions.
- Resolved complex sourcing requirements and reduced carbon emission by approx. 440 metric tons in the past 1 year.
By the numbers
2x eCommerce
Daily Sales
20x increase
in SKUs Exposed
60% reduction
Deployment Cycle Times
Solution Approach
IBM Sterling Order Management containers are delivered as three images — om-base, om-app, and om-agent — through the IBM Entitled Registry using licensed API keys that enable customers with an easier pull access to their local registries or CI/CD pipelines. The deployment charts are readily available in the Red Hat OpenShift Helm Catalog.
om-app — Order Management application server image handling synchronous traffic patterns embedded with IBM WebSphere Liberty application server. This is mainly leveraged for fulfillment options traffic arising out of eCom website to order management systems.
om-agent — Order Management workflow agent and integration server container to handle asynchronous traffic patterns. This orchestrates order fulfillment solution through batch processing of entities.
om-base — Base image provisioned on IBM Cloud Container Registry (Image Registry) and enabled for adding product extensions/customizations to create a customized image, which will be leveraged both by om-app and om-agent process.
IBM DB2 and IBM MQ are deployed outside of the cluster on Azure Virtual machines. The Azure NetApp is used as a Network File Storage (NFS) share which is used for persistent volume storage for deploying the pods and also as mount points to be created for MQ. The latest code should be pulled from the Azure DevOps repository and custom images should be built, which should be pushed to Image Registry as part of the build process either using Jenkins or the Openshift BuildConfig process. Custom images should be deployed to Cluster using the Helm Chart provided by IBM for the OpenShift version and using the Helm deployment process.
Customized application, agent and integration servers will be deployed as pods in the Open Shift cluster. The client will access these pods through Open Shift routes. As far as the deployment strategy goes, the solution has leveraged both the Recreate and the RollingUpgrade option during the Helm upgrade process. The Helm upgrade process ensures a near-zero downtime deployment strategy in Kubernetes. That upgrade process allows for high availability of OMS for eCom fulfillment through the RollingUpgrade option for om-app image, which is a synchronous traffic management system, whereas om-agent uses the Recreate option whereby asynchronous traffic doesn’t create a memory overhead.
Prometheus scripting has been leveraged to raise alerts and notify via email to check the liveliness of any POD in a given namespace in an RHOS Kubernetes cluster. This is a preventive maintenance activity with the help of kube-state-metrics provided by the Kubernetes layer.
Also, CPU threshold-based alerting and POD memory spike alerting has been managed using Prometheus.