I've been testing ignite durable memory with trying to load a lot more data than I have configured for the dataregion, thereby using disk for a lot of data. I was wondering how indexes get persisted to disk in this situation where more data exists than will fit in memory? Is there a way to configure index memory size? The reason I ask this is because I notice that count queries are very slow with disk utilization, but if an index is in memory it should be fast to perform the count. Even select one field queries that utilize an index and return a relatively small data set (like 4000 rows) takes minutes. If the index were in memory, it should be pretty cheap to read from the disk once the items are known. It also seems weird that the same query does not get faster when I run it again, shouldn't those 4000 items be pulled into memory?
-- Sent from: http://apache-ignite-users.70518.x6.nabble.com/