If RF > 1, the consistency level at QUORUM cannot guarantee strict isolation (for normal mutation or batch). If you look at this slide: http://www.slideshare.net/doanduyhai/cassandra-introduction-apache-con-2014-budapest/25, you can see that the mutation is sent by the coordinator, in parallel, to all replicas.
Now it is very possible that due to network latency, the mutation is applied on the first replica and is applied with "some delay" (which can be at the order of microseconds) on other replicas. Theoretically, one client can read updated value on first replica and old value on the other replicas, even at QUORUM. I think that reading at ALL may guarantee to have a "read isolation" in practice but I'm not sure it can be considered "isolated" from the theoretical definition of isolation. On Tue, May 19, 2015 at 7:57 AM, Martin Krasser <krass...@googlemail.com> wrote: > Hi DuyHai, > > thanks for your answer. What if I set RF > 1 and the consistency level for > reads and writes to QUORUM? Would that isolate the single-partition batch > update from reads? (I do not consider node failures here between the write > and the read(s)). > > > On 19.05.15 07:50, DuyHai Doan wrote: > > Hello Martin > > If, and only if you have RF=1, single partition mutations (including > batches) are isolated. > > Otherwise, with RF>1, even a simple UPDATE is not isolated because one > client can read the updated value on one replica and another client reads > the old value on another replica > > > > On Mon, May 18, 2015 at 12:32 PM, Martin Krasser <krass...@googlemail.com> > wrote: > >> Hello, >> >> I have an application that inserts multiple rows within a single >> partition (= all rows share the same partition key) using a BATCH >> statement. Is it possible that other clients can partially read that batch >> or is the batch application isolated i.e. other clients can only read all >> rows of that batch or none of them? >> >> I understand that a BATCH update to multiple partitions is not isolated >> but I'm not sure if this is also the case for a single partition: >> >> - The article Atomic batches in Cassandra 1.2 >> <http://www.datastax.com/dev/blog/atomic-batches-in-cassandra-1-2> says >> that *"... we mean atomic in the database sense that if any part of the >> batch succeeds, all of it will. No other guarantees are implied; in >> particular, there is no isolation"*. >> >> - On the other hand, the CQL BATCH >> <https://cassandra.apache.org/doc/cql3/CQL.html#batchStmt> docs at >> cassandra.apache.org mention that *"* *... the [batch] operations are >> still only isolated within a single partition"* which is a clear >> statement but doesn't it contradict the previous and the next one? >> >> - The CQL BATCH >> <http://docs.datastax.com/en/cql/3.1/cql/cql_reference/batch_r.html> >> docs at docs.datastax.com mention that *"... there is no batch >> isolation. Clients are able to read the first updated rows from the batch, >> while other rows are still being updated on the server. However, >> transactional row updates within a partition key are isolated: clients >> cannot read a partial update"*. Also, what does *"transactional row >> updates"* mean in this context? A lightweight transaction? Something >> else? >> >> Thanks for any help, >> Martin >> >> > > -- > Martin Krasser > > blog: http://krasserm.github.io > code: http://github.com/krasserm > twitter: http://twitter.com/mrt1nz > >
