Hi Josh, > Did you already post the results of: > cat /proc/sys/vm/zone_reclaim_mode
zone_reclaim_mode was set on 0 for all my tests. I’ve also set it to the other values (1, 2, 4) but there was no improvement. Tests results are the following (1 run for each test): echo 0 > /proc/sys/vm/zone_reclaim_mode 130s echo 1 > /proc/sys/vm/zone_reclaim_mode 129s echo 2 > /proc/sys/vm/zone_reclaim_mode 134s echo 4 > /proc/sys/vm/zone_reclaim_mode 131s > Also, how big did you say your dataset is? Based on the output of free, > you're certainly not using all the memory you have. That could be just > because you haven't accessed that much of your dataset, or it could be > because zone reclaim is preventing you from using your entire amount of RAM > as file system cache. The table I use for this test is about 20milion row, has less than 10 columns (a small table) and has 4 indexes. Other tables I use are partitioned and consists of a total of 1.6bilion rows, 757 million rows ad so on descending. Thank you for your help. Best regards, Pietro
