Managed Databases: Relational vs NoSQL in the Cloud
Running your own database means backups, patching, replication, and 3am pages. Managed databases hand all of that to the cloud, so the real question b
Tag archive
Running your own database means backups, patching, replication, and 3am pages. Managed databases hand all of that to the cloud, so the real question b

DynamoDB vs Amazon Aurora compared — NoSQL key-value store versus managed MySQL/PostgreSQL-compatible relational database, from joins to scaling and pricing.

We're excited to announce that LioranDB V2 Pre-Alpha will officially launch on 16 August 2026. This...
Bigtable is the most misunderstood database on Google Cloud. Teams reach for BigQuery when they...
Bigtable ist die am meisten missverstandene Datenbank in Google Cloud. Teams greifen zu BigQuery,...

How we made small, cheap LLMs reliably query DynamoDB — prescriptive tool validators that treat rejection messages as teaching signals, aggressive result caps, and an eval harness that runs real agent loops against DynamoDB Local. The engineering story behind DynoTable's AI assistant.

How we built a local SQL engine for DynamoDB — a database with no JOIN, no GROUP BY, and no COUNT. Streaming pages into embedded SQLite, the parser that silently lied to us, and why we rebuilt the aggregation engine when we realized the competitor is a throwaway script.

Why DynoTable never writes to DynamoDB when you click Save — every edit becomes a reviewable diff in a staging area and ships as a transaction guarded by optimistic locks. The engineering story - the cross-tab bug that reshaped the design, the refactor that deleted a subsystem, and why AI agents made staging the load-bearing wall.

What it took to build a safe DynamoDB MCP server inside a desktop app — an OAuth 2.1 authorization server in Electron, per-connection AWS credential isolation, consent-time region freezing, live license clamping, and MFA completed over MCP elicitation. The engineering story behind DynoTable's MCP server.
People often ask me why I decided to build a database. The answer goes back much further than...

The DynamoDB adjacency list pattern — model a social follow graph as edge items in one table, then traverse both directions of the relationship with a single GSI.

How to delete multiple items in DynamoDB — BatchWriteItem delete loops, TransactWriteItems, PartiQL batches, TTL, and staged multi-row deletes in DynoTable — plus why there is no DELETE WHERE.