Hacker News new | past | comments | ask | show | jobs | submit login
'Slacker DB' shootout: SimpleDB vs. CouchDB vs. App Engine vs. Persevere (infoworld.com)
35 points by snydeq on March 24, 2009 | hide | past | favorite | 5 comments



I don't think the author fully understood the purpose of most of the tools he was dabbling with – he spent a lot more time making jokes about the differences between imaginary age groups than what the tools enable you to do. He talked about CouchDB but said he couldn't think of useful things one could do with reductions outside of counting.. come on. (Plus, you don't even need reduce to count maps to a key in CouchDB.)

Kind of a fluff piece. Lots of buzzwords. I had forgotten about Persevere though, so thanks for the reminder :)


"Now that disk space is so cheap and many of the data models don't benefit as much from normalization, JOINs are easy to leave behind."

JOINs and data normalization were not created to get around disk space constraints.

http://en.wikipedia.org/wiki/Database_normalization#Objectiv...

For the average web application not running on app-engine like infrastructure, I believe the following order of questions would be more prudent:

   1. Do I need a database?
   2. Can I work with a relational database? SQLite? PostgreSQL/MySQL?
   3. Now, do I really need the new hotness?
And if you do need the new hotness, may I also suggest Tokyo Cabinet? I am afraid I can't rate its 'value' on a scale of 10, but I'd say it's pretty high for a free product.


I have been trying out Redis (http://code.google.com/p/redis/) which is also quite good.


It seems like for 1, if you just have a bunch of files somewhere, you should be able to keep that in source control. When is it a good idea (if ever) to keep a SQLite database in git?





Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: