The R+W > RF requirement for strong consistency applies regardless of whether your data is 'immutable' or is being updated. A W=1, R=1 approach will not guarantee consistency between reads and writes.
> R=1 might cassandra look on one of the two nodes, find no data there, and prematurely give up? In essence yes, but Cassandra is not 'prematurely giving up', its doing exactly what you as the client are telling it to do: return once *one* node has been checked for the data you are requesting. > does R apply only to (possibly failed) read attempts or only to successful reads Not quite sure what you are asking but the R+W > RF requirement for strong consistency only applies to successful reads and writes. You will get UnavailableExceptions client side (for reads and writes) if the requested consistency level could not be met. Dan From: Anthony Wilcox [mailto:peacepatr...@hotmail.com] Sent: October-07-11 9:33 To: user@cassandra.apache.org Subject: RE: Immutable CFs and read consistency The last sentence should have been "In other words, does R apply also to (possibly failed) read attempts or only to successful reads?" Anthony _____ From: peacepatr...@hotmail.com To: user@cassandra.apache.org Subject: Immutable CFs and read consistency Date: Fri, 7 Oct 2011 06:21:04 -0700 We have a Column Family that is immutable (no updates after the first write). Suppose we use RF=2 and W=1. Do we still need R=2, so that R+W>RF? Or is it sufficient to have R=1? My guess is yes, R=1 is sufficient since if it reads a row, the row has to be correct: it can't get stale data. But can it get NO data? That is, with R=1 might cassandra look on one of the two nodes, find no data there, and prematurely give up? In other words, does R apply only to (possibly failed) read attempts or only to successful reads? Thanks, Anthony No virus found in this incoming message. Checked by AVG - www.avg.com Version: 9.0.914 / Virus Database: 271.1.1/3942 - Release Date: 10/06/11 14:34:00
