True, but given that replicated caches have readFromBackup = true is that 
significant? Feels like it's a much easier and safer option than trying to 
write your own affinity function.

> On 31 Mar 2020, at 16:28, Evgenii Zhuravlev <[email protected]> wrote:
> 
> Stephen,
> 
> Even for Replicated cache, part of the partitions will be primary on one node 
> and other on another node. So, I think it's possible to create a custom 
> affinity function here.
> 
> Best Regards,
> Evgenii
> 
> вт, 31 мар. 2020 г. в 00:20, Stephen Darlington 
> <[email protected] <mailto:[email protected]>>:
> If you have enough memory to store all your data on a single machine, you can 
> just use a replicated cache.
> 
> Otherwise, you can use a partitioned cache and use “affinity” to make sure 
> the data you need in a single transaction is located on a single machine.
> 
> Regards,
> Stephen
> 
> > On 30 Mar 2020, at 19:17, eerick <[email protected] 
> > <mailto:[email protected]>> wrote:
> > 
> > Dear all,
> > 
> > Is there a way I can configure a 2-node cluster as primary/standby mode,
> > i.e., ALL DATA stores primarily on 1 node and be replicated to backup node.
> > When primary goes down, backup node picks up.
> > 
> > I wanna setup like that because I have a lot of cross-node transactions and
> > I tested that it has higher throughput on single than 2 nodes.
> > 
> > Thanks & Regards,
> > eerick
> > 
> > 
> > 
> > --
> > Sent from: http://apache-ignite-users.70518.x6.nabble.com/ 
> > <http://apache-ignite-users.70518.x6.nabble.com/>


Reply via email to