Just one quick correction that I am alw2ays quick to point out: spanning across logical volumes or LUNs (which are also logical constructs) is useless unless those logical constructs are backed by separate physical disks/spindles. If the logical constructs use the same disks/spindles, you've gained nothing.
J From: [email protected] [mailto:[email protected]] On Behalf Of Mohamed, Hatem (ext) Sent: Monday, December 23, 2013 12:39 PM To: [email protected] Subject: RE: [mssms] SCCM & SQL, making multiple database files >From reading that section of the book I also own, this is presented in an >example where he mentions you can find other best practices out there but for >this example they chose to do one file per core. I'm not going to try to speak >for other people so I will just share what I do know. This is a performance optimizing method which you aren't required to do, but it can help with performance such as handling SQL operations in parallel and also spanning your database across several logical disks and/or LUNS. Most customers are plenty satisfied with spanning the database (mdf), log (ldf), and tempdb across separate RAID volumes which Microsoft mentions: giving the database, log and tempdb their own dedicated RAID volume. (http://technet.microsoft.com/en-us/library/hh846235.aspx#BKMK_ReqDiskSpace) Basically, I would just consider if you actually need to do this or just spanning the database, log and tempdb on separate RAID volumes would be enough. You can still tune the autogrowth settings, as well as setup your memory allocation settings to provide good performance. (there are some autogrowth recommendations from MS in the previously provided link.) From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Stephen Owen Sent: Monday, December 23, 2013 12:51 PM To: [email protected]<mailto:[email protected]> Subject: [mssms] SCCM & SQL, making multiple database files HI all, I've been rereading Kent Agerlund's Mastering the Fundamentals book for Config Man 2012, and one of the things he recommends in the Post-SQL Configuration section on page 32 is that: One of the recommended post installation tasks is to create additional SCCM database files[..] since the server has two CPUs with six cores each, you will create 11 additional files. The estimated size of the database is 30 GB, which will be split into 11 files of 3 GB each that will autogrow 1gb at a time. [PARAPHRASED] The chapter continues on with a script that accomplishes the splitting but doesn't explain why this is done. I've looked around and basically can't figure out what any of this means, and can't find any blog posts or recommendations or (most needed) an explanation of what this is doing and why I'd want to do it. If you guys could shed some light on this, I would appreciate it.
