I am looking to confirm if this behaviour is normal as I got a new server and I am testing some SQL queries and they seem to be slightly slower than on the old machine. This is freebsd 14.1
When I copy a bunch of large files between 2 folders on the same disk, I am getting the following output from zpool iostat
capacity operations bandwidthpool alloc free read write read write---------- ----- ----- ----- ----- ----- -----zroot 90.6G 797G 8 61 272K 1.86Mzroot 90.6G 797G 0 0 0 0zroot 90.6G 797G 0 0 0 0zroot 90.6G 797G 0 125 0 1.12Mzroot 90.6G 797G 0 0 0 0zroot 90.6G 797G 0 0 0 0zroot 90.6G 797G 0 0 0 0zroot 90.6G 797G 0 138 0 2.23Mzroot 90.6G 797G 2.05K 0 138M 0zroot 90.6G 797G 1.66K 0 122M 0zroot 90.6G 797G 1.80K 1.22K 98.4M 91.7Mzroot 91.0G 797G 3.75K 3.37K 189M 249Mzroot 91.0G 797G 3.19K 2.46K 142M 240Mzroot 91.2G 797G 2.53K 593 108M 49.9Mzroot 91.2G 797G 1.84K 0 86.9M 0zroot 91.2G 797G 3.77K 2.81K 181M 268Mzroot 91.2G 797G 2.20K 0 98.5M 0zroot 91.5G 797G 3.27K 3.02K 159M 295Mzroot 91.5G 797G 2.63K 0 144M 0zroot 91.4G 797G 3.52K 3.46K 204M 326Mzroot 91.4G 797G 2.26K 0 128M 0zroot 91.4G 797G 1.85K 1.29K 112M 117Mzroot 91.4G 797G 2.67K 2.75K 118M 244Mzroot 91.4G 797G 1.43K 0 51.6M 0zroot 91.4G 797G 1.84K 873 74.8M 84.0Mzroot 91.3G 797G 2.99K 1.86K 161M 188Mzroot 91.3G 797G 1.40K 0 70.7M 0zroot 91.3G 797G 1.81K 1.03K 79.6M 97.5Mzroot 91.6G 796G 2.84K 2.22K 137M 212Mzroot 91.6G 796G 1.50K 0 81.5M 0zroot 91.6G 796G 1.42K 0 73.9M 0zroot 91.5G 797G 2.94K 3.28K 159M 315Mzroot 91.5G 797G 2.36K 0 121M 0zroot 91.5G 797G 3.80K 3.58K 173M 331Mzroot 91.5G 797G 2.84K 0 123M 0zroot 91.3G 797G 2.74K 2.84K 175M 275Mzroot 91.3G 797G 2.35K 0 108M 0zroot 91.3G 797G 3.86K 3.51K 170M 326Mzroot 91.3G 797G 2.23K 0 93.8M 0zroot 91.5G 796G 3.52K 2.74K 171M 274Mzroot 91.5G 796G 1.93K 0 106M 0zroot 91.5G 796G 1.90K 0 116M 0zroot 91.5G 797G 2.85K 3.90K 188M 347Mzroot 91.5G 797G 1.50K 0 90.2M 0zroot 91.5G 797G 1.97K 0 131M 0zroot 91.5G 796G 3.26K 4.83K 215M 405Mzroot 91.5G 796G 2.53K 0 161M 0zroot 91.6G 796G 3.15K 4.66K 220M 399Mzroot 91.6G 796G 2.65K 0 150M 0zroot 91.4G 797G 3.66K 4.39K 202M 381Mall the zeroes in read and write columns worry me. On old server there are no zeroes and copy speed is consistent, but this new server reaches 189M read speed while the old server is consistent at around 80M. It seems like on the new server reads saturate writes. Is there any setting of zfs (I have default) that I could change to improve this? The disk is some nvme SSD at ovh, a single drive zfs