cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hari Sekhon (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-7877) Tunable consistency across datacenters - LOCAL_QUORUM + 1 at other DC
Date Thu, 04 Sep 2014 11:11:51 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-7877?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Hari Sekhon updated CASSANDRA-7877:
-----------------------------------
    Description: 
Right now tunable consistency allows for things like local quorum and each quorum across multiple
datacenters.

I propose something in between where you achieve local quorum + 1 node at other DC. This would
make sure the data is written to the other datacenter for resilience purposes but be better
performing that having to do an each quorum write at both datacenters. Mind you re-reviewing
each_quorum I'm not sure how much more performant this would be...

EDIT: thinking about this more this is probably reasonably covered by each quorum given that
by the time you write to 1 replica node in other DC you may as well write to 2 in which case
each quorum is probably the way to go.

Although if you have 3 or more datacenters then perhaps there should be an option to do each
quorum in local + quorum in at least 1 other DC before confirming write to prevent data loss
on site failure?

  was:
Right now tunable consistency allows for things like local quorum and each quorum across multiple
datacenters.

I propose something in between where you achieve local quorum + 1 node at other DC. This would
make sure the data is written to the other datacenter for resilience purposes but be better
performing that having to do an each quorum write at both datacenters. Mind you re-reviewing
each_quorum I'm not sure how much more performant this would be...


> Tunable consistency across datacenters - LOCAL_QUORUM + 1 at other DC
> ---------------------------------------------------------------------
>
>                 Key: CASSANDRA-7877
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7877
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Hari Sekhon
>            Priority: Minor
>
> Right now tunable consistency allows for things like local quorum and each quorum across
multiple datacenters.
> I propose something in between where you achieve local quorum + 1 node at other DC. This
would make sure the data is written to the other datacenter for resilience purposes but be
better performing that having to do an each quorum write at both datacenters. Mind you re-reviewing
each_quorum I'm not sure how much more performant this would be...
> EDIT: thinking about this more this is probably reasonably covered by each quorum given
that by the time you write to 1 replica node in other DC you may as well write to 2 in which
case each quorum is probably the way to go.
> Although if you have 3 or more datacenters then perhaps there should be an option to
do each quorum in local + quorum in at least 1 other DC before confirming write to prevent
data loss on site failure?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message