Hi Alex, I am simply using GridGain/Ignite web console to generate a project for testing cacheload of a single table from my persistent storage. For evaluation/testing purposes, I am running a single node in my cluster on a 2CPU 32Gb Windows machine.
For testing, I created 2 projects. Test1 is where the table is imported as a partitioned cache with no (or default) affinity defined. Test2 is where the same table is imported as a partitioned cache with an affinity key for colocating data based on a column that is not part of PK. In case of Test1, I am able to import/load cache with ~537K records in 20 seconds. For Test2, the same data takes a little over 3 minutes. This is just a sample set of my overall data for this table. My total data volume is over 42 million records. I want to partition and colocate my data... but for now trying to understand the reason for performance difference between cache load times. I am attaching the two projects as zip files. This was directly generated from Web console. The only changes I make are correcting maven dependency for mssql jdbc driver (8.4.1.jre14) in pom.xml and putting my persistent DB details in secret.properties. I have reviewed the links you provided and the affinity definition seems to be done correctly in Test2 project as far as I can see. Regards, Gautam Test1Cluster-project.zip <http://apache-ignite-users.70518.x6.nabble.com/file/t3063/Test1Cluster-project.zip> Test2Cluster-project.zip <http://apache-ignite-users.70518.x6.nabble.com/file/t3063/Test2Cluster-project.zip> -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
