[ https://issues.apache.org/jira/browse/CASSANDRA-11521?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15232715#comment-15232715 ]
Brian Hess commented on CASSANDRA-11521: ----------------------------------------- [~slebresne] - I like this new approach better. I think it simplifies things a bit and I'm worried about the server easily overpowering the client, which I think could be really easy to do (then we'd have to think about things like back-pressure, etc). There could be a way to tell the server that the client is going to ask for all (or a lot) of the pages, so keep this stuff ready to flow, etc. Additionally we could have a setting that will tell the server "if you haven't heard me ask for the next page (or given some heartbeat) in a X long, then feel free to clean things up and throw an error if I ask for the next page later", or something, so that we don't have resources tied up even if the client dies. WRT CL, with this approach, I don't quite see why you would have to stick to CL_ONE here. That said, starting with CL_ONE and "growing" to other CL's is probably okay. Just not entirely sure what it gains given this new approach. > Implement streaming for bulk read requests > ------------------------------------------ > > Key: CASSANDRA-11521 > URL: https://issues.apache.org/jira/browse/CASSANDRA-11521 > Project: Cassandra > Issue Type: Sub-task > Components: Local Write-Read Paths > Reporter: Stefania > Assignee: Stefania > Fix For: 3.x > > > Allow clients to stream data from a C* host, bypassing the coordination layer > and eliminating the need to query individual pages one by one. -- This message was sent by Atlassian JIRA (v6.3.4#6332)