In cloud computing environments, data storage systems often rely on optimistic replication to provide good performance to geographically disperse users and to allow operation even in the presence of failures or network partitions. In this scenario, it is important to be able to accurately and efficiently identify updates executed concurrently. In this paper, first we review, and expose problems with current approaches to causality tracking in optimistic replication: these either lose information about causality or do not scale, as they require replicas to maintain information that grows linearly with the number of clients or updates. Then, we propose a novel, scalable solution that fully captures causality: it maintains very concise information that grows linearly only with the number of servers that register updates for a given data element, bounded by the degree of replication. Moreover, causality can be checked in O(1) time instead of O(n) time for version vectors. We have integrated our solution in Riak, and results with realistic benchmarks show that it can use as little as 10% of the space consumed by current version vector implementation, which includes an unsafe pruning mechanism.
- Checks causality in O(1) time instead of O(n) in most cases.
- Multiple concurrent clocks in the same server.
- Allows partial conflict resolution, instead of the "all or nothing" conflict resolution in version vectors.
This implementation is based on the following article: Dotted Version Vectors: Logical Clocks for Optimistic Replication
-
sync(S1, S2): Takes two clock sets and returns a clock set. It returns a set of the more up-to-date concurrent clocks for the union of S1 and S2, while discarding obsolete clocks;
-
update(S, Sr, r): Takes a set of clocks S (the context supplied by the client), the set of clocks Sr in the replica node (for the given key), and the replica node id r, and returns a clock for a new replica. This clock should: 1) dominate all clocks in S, 2) dominate only those clocks in the system that are already dominated by S and 3) not be dominated by any join of clocks in the system.
It exports the following functions:
- fresh/0
- descends/2
- sync/2
- update/3
- equal/2
- increment/2
- merge/1
Article:
- Carlos Baquero cbm@di.uminho.pt
- Paulo Sérgio Almeida psa@di.uminho.pt
- Victor Fonte vff@di.uminho.pt
- Nuno Preguiça nmp@di.fct.unl.pt
- Ricardo Tomé Gonçalves tome@di.uminho.pt
Implementation:
- Ricardo Tomé Gonçalves tome.wave@gmail.com
% Create a new clock
% C = {}
C = dottedvv:fresh(),
% Increment clock C in _Id_
% C2 = {[], {id,1}}
C2 = dottedvv:increment(id, C),
% Get an updated clock in server _Id_, from _arg1_ (client clock) and _arg2_ (server clock)
% C3 = {[{id,1}], {id,2}}
C3 = dottedvv:update(C2, C, id),
% Apply _sync_ with result of _update_ and the server clock, to discard all outdated information
C3 = dottedvv:sync(C3, C).- Add new cover tests