3 ms·
> built around the idea of a write never failing. What am I missing if I think this sounds like a very interesting thought experiment, but completely disconnec
by kqr 8y ago
> built around the idea of a write never failing.
What am I missing if I think this sounds like a very interesting thought experiment, but completely disconnected from reality?
- zzzcpan 8y agoIn distributed systems getting to a failing state is exactly the same as getting to a succeeding state, both have to make CAP consistency/availability trade offs. So in many cases not failing is just easier.
- vorpalhex 8y agoThe tradeoff is "consistency". As long as some node of Mongo is running, it'll do it's absolute damned best to accept a write, including buffering in memory and all kinds of other tricks. However, if you write to it and then immediately read.. you might not get back what you just wrote (but it will eventually show up in reads). Likewise, you might get different responses from different nodes in some weird situations. A few things to note: This weird behavior is only in a failure state (eg a network partition), and I believe it's no longer the default but can be enabled. It's very useful for lots of applications, but isn't the right choice for all applications.