I would like to understand how SQL queries are executed on Ignite Server
Nodes. Each Ignite Server Node has some number of primary partitions it is
responsible for. When a query is sent, let' say through JDBC interface, this
query is routed to all the server nodes where data resides. So the question
is, if a server node has 5 primary partitions it is responsible for, does
the server node spawn multiple threads and process these partitions
concurrently, or does it use a single thread, and process partitions 1 by 1,
keeping temporary results in memory.
If it does process partitions concurrently on a single node, another
question is, if there are more number of partitions than available cores, is
it smart enough to grab all available cores, and then cycle through all
partitions in chunks of available cores keeping temporary results in memory
and merging them as needed.
I am trying to get a sense of the SQL query processing engine and its
internals especially within a single server node.
Sent from: http://apache-ignite-users.70518.x6.nabble.com/