Yes, there is definitely a performance drawback (in terms of latency) and as you said the data might traverse multiple intercontinental links before arriving to you. Multiple nodes (4 if I am correct) will be holding the same chunk so it might be possible to select the fastest amongst them. That won’t guarantee the nodes are going to be close to you though.
For some applications it might not be an issue. We will only find out with large-scale tests how the network actually performs in practice.
Down the road, some optimizations might be possible, such as allowing direct connections between nodes for exchanging public data. Maybe it could be possible for users to pay a maintenance cost for their data using Safecoins that would be proportional to the number of nodes that are holding a copy.
I believe the priority at the moment is to get the behaviour of the network right for holding highly sensitive data securely, such as your Session ID or, let’s say, your bitcoin wallet private keys. For the rest, I believe it will be possible to tune the later versions of the network to be faster, and even maybe allow users to choose a different compromise between privacy and performance.