The architecture behind modern cloud data warehouses
summary
Compares first‑gen cloud warehouses like Redshift (Postgres 8 derivative running on EC2 with attached disks) to newer designs that use separation of compute and storage via object stores (S3/GCS), claiming storage costs drop and capacity becomes “infinite.” Describes MariaDB ColumnStore as an MPP columnar engine that opens many threads to S3, handles billions of rows and petabyte scale, and enables HTAP-style joins by replicating small dimensions. Describes SkySQL on Kubernetes with a ServiceNow portal, per‑customer private clusters, and a $500 trial credit.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:25Basic concepts
- 3:08History
- 4:04Cloud
- 4:34Redshift
- 6:56Redshift architecture
- 7:45Open source data warehousing
- 11:06Distributed architecture
- 12:03Parallelity
- 12:36Realtime
- 17:13Smart engines
- 18:22Query options
- 18:39Joints
- 19:30Hybrid example
- 24:41Sky SQL
- 25:04Sky SQL Architecture
related talks
41mThe New Generation Of Data Stores
32mHow open source is re-shaping the cloud data warehouse landscape
41mOn Data: Sky above, sand below, peace within
55mThe Great Data Debate 2022
28mModernize Your Data Warehouse
43mData warehousing reinvented for today's needs
26mColumnar Storage: Redefining Data Management for the Modern Era
56mUnderstanding and achieving a modern data architecture