Auto scaling Azure Cosmos DB for PostgreSQL with Citus, Grafana & Azure Serverless

on YouTube ↗32m

" summary

Shows a working pipeline that auto-scales an Azure Cosmos DB for PostgreSQL (Citus) cluster using Grafana alerts, Azure Functions (Serverless) calling ARM, and the Postgres client API; demo runs pgbench, adds two workers when CPU >80%, waits ~10 minutes for nodes to provision, then rebalance shards automatically via SQL and pg_dist_node; CPU on workers falls from ~80%+ to under 80 and latency drops from ~150ms to ~100ms; also uses Redis to avoid duplicate runs and isolates a tenant when it uses >50% of a shard.

AI-written summary · about

" chapters (7)

1. 0:00Introduction
2. 3:33Why auto scaling?
3. 4:59Auto scaling a Citus cluster
4. 8:31Grafana
5. 11:54Azure Functions (Serverless)
6. 13:23Azure Resource Manager (ARM)
7. 15:12Demo of Autoscaling Azure Cosmos DB for PostgreSQL

" related talks

• 16mOnline Scale-Out and Cloud Data IngestionAdam Wolk · Azure Cosmos DB: Liftoff 2022
• 52mBuilding Distributed OLTP with Azure Cosmos DB for PostgreSQLCharles Feddersen · SQLBits 2023
• 02h 02mAzure Cosmos DB: LiftoffKirill Gavrylyuk, Estefani Arroyo, Iria Osara, Sai Krishna Srirampur, Claire Giordano · Azure Cosmos DB: Liftoff 2022
• 06mKeynote: Azure Cosmos DB for PostgreSQLKirill Gavrylyuk · Azure Cosmos DB: Liftoff 2022
• 44mMulti-tenant SaaS apps made simple on Azure Cosmos DB for PostgreSQLAdam Wolk, Alicja Kucharczyk · SQLBits 2023
• 50mUnleashing the Power of Azure for PostgreSQLShriram M · PGConf India 2025
• 08mFoundations of Distributed PostgreSQL with CitusOnder Kalaci · Azure Cosmos DB: Liftoff 2022
• 13mHow to do Data Modeling for Distributed Postgres in Azure Cosmos DBSai Krishna Srirampur · Azure Cosmos DB: Liftoff 2022