On 10/06/2016 06:02 PM, Dimitri Maziuk wrote: > On 10/06/2016 09:26 AM, Klaus Wenninger wrote: > >> Usually one - at least me so far - would rather think that having >> the awareness of redundany/cluster as high up as possible in the >> protocol/application-stack would open up possibilities for more >> appropriate reactions. > The obvious counter-example is a hard disk failure: they're common on > commodity spinning rust drives and they're cheap and easy to handle at > lower level by throwing in a 2nd one in mdadm raid-1. Not so sure if there wouldn't be fancy things a database or something even higher level could do when it knows that it is sending records to parallel storages. Especially when you think of a clustered service sitting on top ... But it is convenient because all layers on top can be completely agnostic of the duplicity. > > > > _______________________________________________ > Users mailing list: [email protected] > http://clusterlabs.org/mailman/listinfo/users > > Project Home: http://www.clusterlabs.org > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf > Bugs: http://bugs.clusterlabs.org
_______________________________________________ Users mailing list: [email protected] http://clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org
