3 ms·
Lack of UX & no ACID compliance so I wouldn't use it for anything in production.
by mrmondo 12y ago
Lack of UX & no ACID compliance so I wouldn't use it for anything in production.
- aikah 12y agowhat do you mean no ACID compliance ? the db used? And can you be more specific about what's wrong with the UX?
- mrmondo 12y agoCorrect, regardless of what redis tries to sell itself as it is not a database I would trust my data with long term. As far as UX goes the demo site was incredibly hard to use, it doesn't lead your eyes trough a path of workflow, not to mention that narrow vertical columns are a highly inefficient way to digest data.
- cwyers 12y agoIs an online forum an application where you have to worry about trusting your database? If you lose a forum post now and then, it's not a good thing, certainly, but depending on the forum, it could be an acceptable tradeoff.
- Argorak 12y agoConsidering that a forum is also not the most high-traffic kind of application, I do not see the reason why you should use a high-performance K/V-store (with all the tradeoffs coming with it).
- eurleif 12y agoRedis is fully durable if you use append-only files with the "appendfsync always" option set. http://oldblog.antirez.com/post/redis-persistence-demystified.html http://oldblog.antirez.com/post/redis-persistence-demystifie...
- Argorak 12y agoDurable != ACID You can still create an inconsistent state in Redis easily, even it persists well.
- eurleif 12y agoAs far as I know, consistency in ACID just means that the database enforces any constraints that have been defined. Since redis doesn't allow constraints to be defined, doesn't that technically make it consistent, in a vacuous sort of way? You could also view redis's data types as a form of constraint. Using a set instead of a list achieves the same purpose as using UNIQUE in SQL.
- taude 12y agoI was curious about the use of Redis as a permanent data store, too. Is this a common use case for people? I've only used it for a data cache, where the actual values are backed by a relational data store and some logic that calculates expensive numbers.