You know this answer Web... as a consultant, the right answer to any technical question is "it depends".
What kind of storage is it? SAN? NAS? JBOD? SSD? RAID? FC? What's the I/O pattern? Serial? Random? What's the IOPS requirement vs. the IOPS available? What's the real IOPS load? (you probably won't be able to get this one.) What's the backup plan? Or the replication plan? Or the availability group plan? All of the above? What is the organization's sensitivity to downtime? So... back in "the day", with small disks, you absolutely should separate everything: OS on one disk (or RAID-1, the I/O pattern is more or less random with a real hot-spot on the paging file), Data on another disk (or RAID-1, data is random), and Logs on the third disk (or RAID-1, I/O pattern is serial and the blocks are large). I wouldn't ever do that today. I'm gonna put two SSDs on an expensive RAID-1 controller and backup to the cloud (private or public or hybrid is irrelevant). Nobody cares about the I/O pattern on SSD and in this case, adding logical volumes to the physical RAID just complicates life for very little real advantage (there is a potential real advantage if you have an application go amuck and begin to generate TBs of log files - but you should have monitoring as well). From: [email protected] [mailto:[email protected]] On Behalf Of Webster Sent: Monday, June 26, 2017 12:30 PM To: NT Issues ([email protected]) Subject: [NTSysADM] Does Separating Data and Log Files Make Your Server More Reliable? I had always been told to separate everything in SQL Server. https://www.brentozar.com/archive/2017/06/separating-data-log-files-make-server-reliable/ Webster
