With total order and sync replication it is possible for the following: tm.begin() cache(0).put(k,v); tm.commit(); assert cache(0).get(k).equals(v); //always passes assert cache(1).get(k).equals(v); //might fail! //..but eventually(cache(1).get(k).equals(v)) //this will always pass In other words, after a transaction is completed, you are only guaranteed to see its results on the originator node. On replicas the value is only eventually made available. Whilst this is fine for embedded caches, as in practice the replication would happen between different JVMs, this is still relevant for: - hotrod: a client might tx-write on a node and read from the other - tests Cheers Mircea -- Mircea Markus twitter.com/mirceamarkus Sr. Software Engineer, Infinispan http://www.infinispan.org _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev

> if I recall correctly currently (with 2PC) the commit is sent > asynchronously (without waiting to gather acks). The commit used to be send async (default) before 5.2, but now it is sent sync. *But* there were no complains with async commits AFAIK, which is a good thing. _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev

BTW, the results posted, both configurations had the flag syncCommit/RollbackPhase set to false. On 3/7/12 1:41 PM, Paolo Romano wrote: > I did not know about this change, thanks for pointing it out! By the > way, is it still possible to send commit messages asynchronously with > 2PC by changing the default config? I am asking as, in order to ensure > fairness when comparing the performance of the two protocols, it would > be better to make sure that they enforce the same semantic. > > Another possibility would be to extend the TOB-based replication > mechanism to support this semantic. Conceptually, this is not too > complicated, but coding that will still take some time... and I would > prefer to invest effort on TOM+partial replication. What do you think? > > Cheers, > > Paolo > > On 3/7/12 1:17 PM, Mircea Markus wrote: >>> if I recall correctly currently (with 2PC) the commit is sent >>> asynchronously (without waiting to gather acks). >> The commit used to be send async (default) before 5.2, but now it is sent sync. >> *But* there were no complains with async commits AFAIK, which is a good thing. >> >> >> _______________________________________________ >> infinispan-dev mailing list >> infinispan-dev(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/infinispan-dev > _______________________________________________ > infinispan-dev mailing list > infinispan-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/infinispan-dev _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev