Case 5: Inventory Avalanche: Scaling Real-Time Inventory Feed
You joined a large ecommerce company that also has brick-n-motor stores spread around the country. Each store chain updates its stock status of various items to the cloud system in real-time from its pos system and eod-status as batches from its MIS. The updates are in the format `StoreID, SKU | Event | Available Quantity`, where Event could be one of Sold, Damaged, Lost, or Expired.
The brand has become so popular that a couple of new brick-n-motor stores were opened every week in the past year and now the inventory management system (IMS) deployed in cloud that is to listen to the live stock updates from each of the retail stores is chocking for the scale of data ingestion of real-time data feed. The cloud system is having this challenge in spite of leveraging Kafka as message broker for receiving real-time data streams.
As an experienced engineer and architect, where do you see the bottleneck and what do you think could be the possible reasons? And how would you go about addressing this issue keeping in mind the short-term and long-term scaling needs?
As a seasoned engineering manager how would you deal with the different stakeholders involved who keep pinging you in various channels from whatsapp, slack, emails to calls and meetings?