-
Notifications
You must be signed in to change notification settings - Fork 0
Databases
richoH edited this page Apr 27, 2012
·
5 revisions
This app will obviously need a datastore. I'd sooner gauge out my own eyes than using my-fucking-sql on another project, but with that said, add pros and cons as you see fit. I'm divvying this into two seperate tables for ephemeral and permanent storage
- Shallow learning curve for new devs
- Well supported by the community
- Easy to set up replication (in the short term, at least in the number of mirrors we're likely to see)
- Is only a true RDBMS with 3rd party addons
- Only recently managed ACID compliance (?!)
- Serious performance issues
- Limited stored procedures support
- Search is only possible via 3rd parties (Sphinx)
- Indexes are shallow and stupid
- Stupid fast
- Easy to extend at the database level
- Massive stored procedure support (pl/python, pl/sql, pl/PHP, pl/Ruby)
- Smart query planner
- Great debugger
- Steeper learning curve
- Tuning isn't trivial
- Replication is still sketchy (This may be out of date, I seem to remember TL making big strides before 9.1)
are we supposed to put our thoughts here? if so: (otherwise delete this)
- NanoyMaster - I'm preferable to MySQL, yes I do realise the benefits to others, but it is already known, both for current devs and new people that are likley to join and well just easier.
- Internal locking mechanisms are fantastic and eliminate a lot of common coder errors
- Resque is a really mature project for building worker queues
- Reasonbly high overhead
- Can be a pain to debug
- (redis is a ruby thing which means marshalling jobs in such a way that ruby can parse them)
- Stupid fast
- Has no abstraction (Can also be viewed as a plus if we want to do something cool)