Исходное сообщение
"Релиз FreeNAS 0.7, дистрибутива для организации работы сетев..." Отправлено sHaggY_caT, 07-Ноя-09 15:42
>>чтение в gmirror и mdraid на зеркале > >хоть мне сия фраза и не понятна, но не все же не >могу не согласится, что скорость чтения возрастает. а если еще правильно >выбран balance (round-robin,load,prefer,split), то можно добиться очень хороших результатов. Да, написала немного не грамотно, но я и не пыталась умничать. Для оппонентов Вот про md: (http://tldp.org/HOWTO/Software-RAID-0.4x-HOWTO-8.html) Linux MD RAID-1 (mirroring) read performance: MD implements read balancing. That is, the RAID-1 code will alternate between each of the (two or more) disks in the mirror, making alternate reads to each. In a low-I/O situation, this won't change performance at all: you will have to wait for one disk to complete the read. But, with two disks in a high-I/O environment, this could as much as double the read performance, since reads can be issued to each of the disks in parallel. For N disks in the mirror, this could improve performance N-fold. Надеюсь, для оппонентов tldp.org подойдет в качестве авторитетного источника :)? А пзомерки, повторюсь, есть в гугле и в избытке... Вот man gmirror:   -b balance  Specifies balance algorithm to use, one of:                              load         Read from the component with the                                           lowest load.                              prefer       Read from the component with the                                           biggest priority.                              round-robin  Use round-robin algorithm when                                           choosing component to read.                              split        Split read requests, which are big-                                           ger than or equal to slice size on N                                           pieces, where N is the number of                                           active components.  This is the                                           default balance algorithm.