"It turns out that this one little decision makes it so that we can't horizontally scale that layer of our architecture without losing all the data already there (because all of the keys would point to the wrong server if we added a new one)."
I'm surprised they're looking to switch stacks entirely as opposed to a consistent hashing / distributed hash table (a la chord, dynamo, etc.)
We can't just switch hashing at this point, because none of the data would be in the right place.
Since we have to move all the data anyway, we figured now would be a good time to switch stacks. Memcachedb isn't really a very solid product, so even if we scale it, it isn't a good long term solution.
Since we have to move all the data anyway, we figured now would be a good time to switch stacks. Memcachedb isn't really a very solid product, so even if we scale it, it isn't a good long term solution.