Thanks Andrew for the explanation. Just one additional question. Lets say i have an apache ignite cluster running on 2 different machines, and we give noOfBackups of a replicated cache as 1, each of the 2 machines will hold 1/2 of the primary partitions and 1/2 of the backup partition.
Currently, each SQL query runs on the primary partitions only, so that a query will be sent to 2 machines and the results will be merged. However in the above case, machine 1 has 1/2 of the primary partition and 1/2 of the backup partition of the machine 2(so that in a way, it holds all the data itself). Theoretically wouldn't it be faster if the query would hit only 1 node/machine since all the data would be on both the machines(in the above case). That was the rationale behind trying to see if the read from backups would help in case of SQL queries. May be, my understanding of this is incorrect. Thanks Tuco -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Basic-question-on-scaling-of-reads-with-ignite-sql-queries-tp11975p12232.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
