These are chat archives for RBMHTechnology/eventuate

20th
Apr 2016
gabrielgiussi
@gabrielgiussi
Apr 20 2016 12:05
Thanks to both of you.
gabrielgiussi
@gabrielgiussi
Apr 20 2016 12:53

@krasserm My last question for now.

Eventuate read/writes to a Cassandra cluster with strong consistency (QUORUM reads/writes)

Each location only needs to persist an event to his local event log in order to say that has been persisted successfully? Then it send quorum writes to Cassandra asynchronously in batchs?
So an event log is independent of the storage backend? Is persisted in the filesystem?

Martin Krasser
@krasserm
Apr 20 2016 13:10
@gabrielgiussi The storage backend for an event log is either Cassandra or Leveldb. In case of Cassandra, each location has its own Cassandra cluster. An event is successfully written to a Cassandra-backed event log if that event has been successfully written to a quorum of replica nodes in the Cassandra cluster.
You can also connect locations with a LevelDB storage backend with those with a Cassandra storage backend.
gabrielgiussi
@gabrielgiussi
Apr 20 2016 13:36
@krasserm So, although Eventuate is AP and allows writes under replication endpoints partitions, when we try to persist an event it may fail because there is no Cassandra QUORUM?
Martin Krasser
@krasserm
Apr 20 2016 13:47
yes, how else would you achieve sequential writes (= event log) at a given location? You may also choose to write to a LevelDB instance i.e. to the local filesystem with the consequence of weaker durability guarantees. AP in Eventuate refers to locations, not the storage backends themselves. See also https://krasserm.github.io/2015/05/25/akka-persistence-eventuate-comparison/ for more details (section event logs).
gabrielgiussi
@gabrielgiussi
Apr 20 2016 13:53
Thanks for your time. I will keep investigating and testing eventuate.
Martin Krasser
@krasserm
Apr 20 2016 17:11
@gabrielgiussi you're welcome, glad to hear