TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

In defense of extreme database-centric architecture [pdf]

3 pointsby steve-chavezover 1 year ago

1 comment

bob1029over 1 year ago
An elegant concept. If your schema is good, this works. Even if it sucks, it will probably still work.<p>We don&#x27;t have the discipline to go all the way, but we are pretty close conceptually - the UI&#x2F;logic layer is still done in code. 100% of the data lives in exactly 1 SQL database, and all of our products are expected to communicate by way of the schema for integration purposes. There are no API contracts or JSON&#x2F;gRPC-serialized things anymore. The schema is the contract and all application state transitions must pass through the transaction log.<p>One database can scale like bananas with the technology we have today. There&#x27;s really no reason to optimize prematurely on this one. Even if your business is so data intensive that one SQL database won&#x27;t work (doubtful), then you still need a place to store <i>metadata</i> regarding the data that presumably won&#x27;t fit so that it can be retrieved by other means. Once we get down it - If you combine a SQL database w&#x2F; some blob storage provider (i.e. S3), what excuse remains from a purely data&#x2F;information theory perspective?<p>I think the dark secret truth that underlies why we don&#x27;t see more of this architecture is simply because it is boring and &quot;too easy&quot;. There&#x27;s not a lot of opportunity to look like a smart-ass in front of others when we are mostly talking about glorified excel documents.