Building the future of time series data using Azure Cosmos DB
summary
Proposes using Cosmos DB (Cassandra API) as a global, managed time-series store for IoT, handling ~5,000 edge devices and about 2 million observations per minute with messages ~8–10 KB containing 10–50 measurements. Implements a tenant-per-table model with timestamp clustering to return latest records; moves RU allocation from tenant to keyspace and shifts to manual scaling for predictable load, observing ~140k RUs used and P95 write latency <10 ms (P99 <20 ms); archives historic data to blob storage.
AI-written summary · about
chapters (6)
- 0:00Intro
- 1:11Introduction to the Use Case
- 4:04Why Cosmos DB?
- 5:57Data model design
- 8:42Challenges & optimizations
- 9:50Outcomes & Key considerations
related talks
10mBuilding IoT Apps with JSONB & Timeseries
46mDesign and build multi-tenant SaaS apps at scale with Azure Cosmos DB
45mGetting the Most from NoSQL in the Cloud with Azure Cosmos DB
02h 02mAzure Cosmos DB: Liftoff
01h 02mAzure Cosmos DB - Low Latency and High Availability at Planet Scale
39mBuilding Event-Driven Systems at Scale with Cosmos DB
21mBuild Scalable Inventory System with Azure Cosmos DB
01mCustomer Story: Empowering Health Research with Azure Cosmos DB