Roundtable: Cloud Databases as a Service
summary
Roundtable reports concrete choices and outcomes: reusing Cassandra and RocksDB components accelerates product cycles, a private cloud runs roughly 5,000 machines to automate recovery and cut manual ops, a two‑person effort ships a transaction layer on Cassandra in ~6 months, speculative retries mask GC/network tail latency, teams prefer JVM for faster development over C++ tradeoffs, LSM trees ease write scaling vs B‑trees, network disks are rejected for low‑latency DBs and automation lets them double machine utilization.
AI-written summary · about
chapters (7)
- 0:00Welcome
- 1:03Speakers introduction
- 14:05Building a DB from scratch
- 29:54Assembling DB from components
- 40:10Why vendors go cloud
- 57:02Sharding vs shared disk
- 1:03:57Summary and discussion
related talks
40mLatency, Throughput & Fault Tolerance: Building a Modern Database
16mBuild for real time on AWS: Thrive in an unpredictable market
50mThe Great Open Cloud Shootout
35mA Customer’s Journey: Migrating from NoSQL to a Cloud Native Distributed SQL Database
57mMulti Tenant NoSQL and NewSQL Cloud data Design Patterns Deep dive
25mData in the Cloud
40mCloud Databases: Pros and Cons
33mBuilding an Elastic, Resilient Data Layer to Power New Retail Experiences