Hi,

I would like to understand which is the best solution to simulate a
transactional "update x where y" logic.

>From what I understand the transaction coordinator has to know all the
modified rows. 
Is there a way to avoid that and have the entire logic (reading filtered
data + updates) locally on all nodes which stores data for a cache?  
In this scenario the transaction coordinator is not aware of the impacted
rows.

Thanks,
Ionut



--
View this message in context: 
http://apache-ignite-users.70518.x6.nabble.com/Update-performance-tp6214.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Reply via email to