Hacker News new | ask | show | jobs
by bradhe 3990 days ago
> * Key Value Stores Redis / Memcache

> * No-SQL land: MongoDB/Cassandra etc.

> * “Traditional” Schema’d SQL: Oracle/Postgres/MySQL/Amazon RDS Etc.

> * Hosted DBs: AWS RDS / DynamoDB / AppEngine Datastore

> * Heavy lifting: Amazon MR / Hadoop (Hive/Pig) / Cloudera / Google Big Query

> * Batshit crazy stuff: Erlang’s Mnesia, iOS’s Core Data

Cringeworthy to be sure.
1 comments
zgoldberg 3990 days ago
Do you disagree with the point?
link
bradhe 3990 days ago
1. "No-SQL land" - Most of those things are NoSQL databases? 2. ""Traditional" Schema'd SQL" - Amazon RDS isn't a database, and those are relational databases. 3. DynamoDB has effectively the same data model as Cassandra. 4. "Hadoop" isn't a database. Cloudera is a database vendor. EMR (what you call Amazon MR?) is Hadoop as a Service. 5. I don't even know what to say about the "batship crazy" thing.

Every data store has a model. Typically we classify them based on their model.

link
zgoldberg 3989 days ago
ah, ok, I understand your complaint. I actually agree with you, I was definitely playing fast and loose with classifications there. That said, my point wasn't about how to classify data storage systems, so much as to emphasize the fact that there are many options and that the architect should be thinking really hard about existing options before rolling their own.
link