http://disqus.com/by/greglinwood/enWed, 22 Jul 2009 08:44:27 -0000http://jasonmassie.com/archive/2009/07/large-pages-and-sql-server/#comment-13105366<p>Sure - but I'm just pointing out that I think it's not a great idea to leave write-access open to kernel memory, purely for the minor perf improvements large pages might bring. If system stability isn't enough of a reason, security should be.</p>greglinwoodWed, 22 Jul 2009 08:44:27 -0000http://jasonmassie.com/archive/2009/07/large-pages-and-sql-server/#comment-13004581<p>You want to be 100% sure of your device drivers before using large pages b/c whilst it's a perf improvement option, it also increases the ability for device drivers to corrupt the o/s (larger pages are marked as read / write protection vs smaller pages enabling kernel code to be marked as read only protection exclusively).</p><p>Hence, it's definitely good for benchmarking on very carefully controlled hardware / drivers, but not good for general population use "in the wild" on commodity hardware.</p>greglinwoodTue, 21 Jul 2009 13:14:07 -0000