The SaaS version of IBM Sterling OMS facilitates real-time order and inventory management for many brands that are household names in the US and internationally, from massive online marketplaces to popular shops for fashion apparel and children’s gifts. For these companies, if the OMS system falters, even for a minute, it is money lost; during Black Friday or other seasonal peaks, it could be millions.

The Sterling OMS team at IBM understands what is at stake, and it commits to a service-level agreement (SLA) of 99.9999% uptime for customers. But making good on this SLA is no simple task.

IBM Sterling OMS SaaS technology stack

• Database components: IBM® Db2®, NoSQL, Cassandra, MongoDB, Elastic Search
• Messaging: IBM MQ, Kafka
• Software defined networking (SDN): istio Service Mesh
• Apache Spark and Apache Hadoop to support AI workloads
• 20+ IBM Cloud® Kubernetes Service clusters, 900+ worker nodes, 60,000 active containers
• Coming soon: Large Language Model (LLM) AI use cases to enhance services for call centers

The SaaS OMS solution hosts more than 300 client environments and integrates them with a diverse technology stack. “Each customer environment is both complex and unique,” says Pradeep Nanjundaswamy, Director of Order Management Development at IBM. “For any implementation, there is a minimum of around 20 integrations with external systems such as payment systems, ERP and delivery services.” Monitoring performance across the whole heterogeneous stack is a must; each component is critical to overall high availability.

Earlier in 2023, the Sterling OMS team had replaced its previous, third-party performance-monitoring solution with IBM Instana Observability.

Black Friday would be the first seasonal surge for the new setup. But from the time they began testing Instana, the OMS team saw advantages they knew would serve them well during the holidays. “The previous solution required many custom integrations that we have been able to replace with out-of-the-box integrations from Instana,” says Shoeb Bihari, Technical Lead / Site Reliability Engineer Advisor on the Sterling team.

The user interface is also better now. “With the other solution, we had to embed its screens into our self-service tool,” explains Nanjundaswamy. “With Instana we rewrote the user screens leveraging Instana APIs, and the user experience has greatly improved.”

The OMS support team now uses Instana to monitor system health and performance in real time and intervene according to the SLAs. “We have browser-based monitoring, covering the application level and infrastructure,” says Dinup Pillai, Senior Technical Staff Member on the Sterling team. This includes synthetic monitoring (also known as active monitoring), which simulates end-user actions from different locations and continually monitors availability and response time. “When any performance issues arise, automated alerts help minimize response time,” says Pillai.

The team also takes advantage of Instana’s flexibility to create personalized dashboards and alerts for API-level and business-level metrics. They can provide role-based visibility to client and internal personnel. Clients get real-time views of business performance, including order volumes and capture rates.

When Black Friday arrived, clients’ order volumes exploded by 20-30 times their normal rates for a solid week. In 2022, over 6 billion API calls¹ hit the OMS system during Thanksgiving week.  A year later, there were about 15 billion API calls—an increase of 120% over the previous year.

So, how did the system do?

Most of the API calls had a response time in sub-seconds. “It was the first year we faced the holiday season with Instana as the monitoring system,” says Nanjundaswamy. “We had 15 billion API calls in 2023 without experiencing severity one incidents or escalations. Overall, it was smooth.”