How Delhivery manages to bill over 10 Mn shipments in real-time
summary
Delhivery explains a Part Truck Load billing system built on OpenSearch that handles over 10M document updates a day and roughly 20+ updates per package by storing package docs, running charge-calculation queries, and writing back cost fields. Scale fixes include shard reconfiguration across a 3-AZ, six‑data‑node cluster and application-level batching of rapid updates into larger bulk writes, which cuts request volume and yields a >70% indexing latency reduction while serving thousands of finance reports and preventing billing disputes for lost or overweight shipments.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:21Delivery scale overview
- 1:05Delivery logistics
- 2:09Financial applications and real-time charge calculation
- 3:13Real-time updates importance
- 4:16Customer benefits of real-time billing
- 4:59OpenSearch in shipping costs
- 6:02Scale challenges & indexing latency
- 7:04Sharding & indexing strategy
- 8:29Optimization results
- 9:13Infrastructure improvements
- 9:34Closing
related talks
05h 34mData @Scale 2024
10mBuilding High-Performance Financial Data Pipelines
48mHyper Speed! When Big Data Blooms
44mScaling OpenSearch at Uber: Operating Resilient, Fault-Tolerant, High-Performance Clus...
40mBilling the cloud: Real world stream processing
41mCapacity and efficiency problems in real-time logistics - and how data at Swiggy solves this
48mHow smava boosts business agility and innovation w/Amazon Redshift
24mHow To Remove Data Siloes: Tips & Considerations