3 ms·
Concurrent Revisions [1] are even better IMO. Much simpler to reason about. In fact, if you know how to work with git and other distributed SCM tools, then you
by naasking 10y ago
Concurrent Revisions [1] are even better IMO. Much simpler to reason about. In fact, if you know how to work with git and other distributed SCM tools, then you already understand the model: branch, do your work locally, then deterministic merge.
[1] http://research.microsoft.com/en-us/projects/revisions/ http://research.microsoft.com/en-us/projects/revisions/
- Jweb_Guru 10y agoYes, I just found out about these recently and am surprised no datastores have been built around them. They are basically a usable version of fork-join causal consistency. They also subsume CRDTs.
- naasking 10y agoDo they? The papers on CRDTs I've skimmed often mention concurrent revisions, but they've never claimed CR subsume CRDTs.
- Jweb_Guru 10y agoThis paper does (in fact, it points out that any total function can be emulated by a query-update automaton, which can be used with concurrent revisions). At least, that's my recollection, it could be a generalization of CR.
- naasking 10y agoWhat is the "this paper" you're referring to? The papers at the link I provided don't mention CRDTs much.
- Jweb_Guru 10y agohttp://research.microsoft.com/pubs/155638/msr-tr-2011-117%282%29.pdf http://research.microsoft.com/pubs/155638/msr-tr-2011-117%28...: "Our definition is strictly more general than CRDTs [17] in the following sense: From any state-based CRDT we can obtain a FJ-QUA by using the same state and initial state, the same query and update functions, a fork function that creates a new replica and then merges the forker state, and a join function that uses the merge. Note that the definition of strong eventual consistency in [17], just like ours, requires that updates can be applied to any state."
- jacques_chester 10y ago> Note that the definition of strong eventual consistency in [17], just like ours, requires that updates can be applied to any state. My favourite gotcha. It solves distributed statefulness! ... so long as someone else solves statefulness!
- naasking 10y agoI don't think that's a fair summary. Stateful cells have non-local and non-commutative evolution, which is why they're hard to distribute. These frameworks eliminate the non-locality, which isn't negotiable if you want to scale, but at the expense of preserving the requirement that you manually ensure commutativity. Not a bad tradeoff, and I highly recommend the CR papers if you haven't read them. I'd say most programs have a natural expression in terms of revisions, and you're left with only a small set of circumstances where state simply must evolve consistently and atomically (airplane seat reservations, for instance). I would still like a framework for reasoning about these circumstances that integrates naturally with CR though.
- jacques_chester 10y agoI hear you, I do. But it irks me when people wave this stuff around like a magic wand, because it's just not. It's like saying you have an O(1) solution to all TSP graphs with the teeny, tiny caveat that you start with a solution to the TSP graph you're working on.
- naasking 10y agoI still think your last statement is unfair. At worst, the caveat is that the TSP solution is for any graph within a certain class of topologies. Concurrent revisions in particular has been implemented to work in concurrent, parallel and incremental computing, with no change in semantics. So it's pretty darn close to a magic wand, but it can't violate the laws of logic.
- Jweb_Guru 10y ago