This is very interesting research, both as a piece of distributed systems theory and as something that's likely to change the way (some) large-scale database systems are designed. When reading this paper it's worth noting that a lot of the conversation, in both academia and industry, around distributed systems is a conversation about constraints. From time to time new and profound results do show up, but the general thread of research is looking at application requirements and looking at system models, and trying to find new and interesting ways to meet these application requirements in the provided system models. This research, the earlier HAT research from the same group, and active research into CRDTs are all great examples of this. I'm not trying to diminish the importance of this research by explaining it this way. Indeed, the 'how do we solve this constraint in this model' research is generally more immediately useful, more applicable, and has a faster turn-around to production systems than the more purely mathematically-focused research in the same fields.
This kind of thing is really going to make it to a database near you, and give distributed databases a great shot at beating single-machine databases in some areas where they lag behind today.
This is very interesting research, both as a piece of distributed systems theory and as something that's likely to change the way (some) large-scale database systems are designed. When reading this paper it's worth noting that a lot of the conversation, in both academia and industry, around distributed systems is a conversation about constraints. From time to time new and profound results do show up, but the general thread of research is looking at application requirements and looking at system models, and trying to find new and interesting ways to meet these application requirements in the provided system models. This research, the earlier HAT research from the same group, and active research into CRDTs are all great examples of this. I'm not trying to diminish the importance of this research by explaining it this way. Indeed, the 'how do we solve this constraint in this model' research is generally more immediately useful, more applicable, and has a faster turn-around to production systems than the more purely mathematically-focused research in the same fields.
This kind of thing is really going to make it to a database near you, and give distributed databases a great shot at beating single-machine databases in some areas where they lag behind today.