Arthi, I found out that there was an issue with visor cmd that was fixed recently https://issues.apache.org/jira/browse/IGNITE-2726 The fix will be available in the next release.
As per SQL performance, the offheap data is deserialized and copied to Java heap when you need to return it as a part of a query result set. In fact SQL is able to look over offheap data without moving it to Java heap.
Makes sense? Regards, Denis On 3/30/2016 8:22 AM, arthi wrote:
Thanks Denis. I could use cache.size(CachePeekMode.OFFHEAP) to the off heap entries count. The visor tool does not show the offheap entries though (I used cache -a to print the detailed stats). Just a quick check on the SQL performance. will there be noticeable difference in SQL join query runtimes when using on heap and off heap mode caches? I did read other posts and documentation that say off heap could be 20% slower or so - is it due to additional de-serializations that may be required, as I understand that entries off heap are serialized. Thanks for all your help. Arthi -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Cache-loading-errors-out-tp3636p3759.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
