Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
sqldeverloper100
Starting Member
3 Posts |
 Posted - 2006-11-14 : 08:01:32
|
| We have a production database (SQL Server SP1) of 15GB size.I was wondering what is the best: To use RAID Stripping or Filegroup Stripping?Or both?First Choice:If I use RAID1+0, I can have 4 physical disks ( 2 for stripping e.g. C: D: and 2 for mirroring lets say C-Mirror: D-Mirror: ) but I could see only one logical disk lets say L:. The RAID mechanism will break my datafiles ( and also every file of my server -operating system files also, I suppose-) to C: and D:.Second ChoiceIf I dont use RAID1+0 but RAID-0 then I can have two seperate RAID-0 devices e.g. 1. RAID-0 C: and C-Mirror and I could see one logical device , lets say L12. RAID-0 D: and D-Mirror and I could see one logical device , lets say L2Now I can save my operating system to L1 and I can create a filegroup FG1 with two files, lets say L1:\myDb1.ndf and L2:\myDB2.ndf. In this case, sql server will choose how to make the stripping of objects belonging to FG1 since the files L1:\myDb1.ndf and L2:\myDB2.ndf are saved in different physical disksWhat is the best: RAID1+0 to choose about the stripping or SQL Server 2005? What if I could combine the two solutions ? If I had RAID0+1 (only one logical disk - see First Choice) then, could the creation of a filegroup help me or it will cause many I/Os because of file fragmentation?Thanks |
|
|
Page47
Master Smack Fu Yak Hacker
2878 Posts |
Posted - 2006-11-14 : 08:11:59
|
| How do you measure best? Do you mean most performant? Do you mean most fault-tolerant?Filegroups allow you to do some cool things like streamline your backup/recovery plan, seperate some heavily used object (like many a covering index or something) onto its own device, move archived data to its own cheaper/slower device, etc. Are these the types of things you consider "best"?Jay White |
 |
|
|
sqldeverloper100
Starting Member
3 Posts |
Posted - 2006-11-14 : 11:08:23
|
Which of the two solutions is the best in terms of I/O performance, sql server response time, time needed for DML statements?Let s say that I could use filegroup in order to put there a very large database. In this case filegroup in different disks (Second choice: SQL Puts the records on different disks) is better than stripping performed automatically by RAID(first choice)?Thanksquote:
Originally posted by Page47
How do you measure best? Do you mean most performant? Do you mean most fault-tolerant?Filegroups allow you to do some cool things like streamline your backup/recovery plan, seperate some heavily used object (like many a covering index or something) onto its own device, move archived data to its own cheaper/slower device, etc. Are these the types of things you consider "best"?Jay White
|
|
|
|
|
|
|
|
|
RAID Stripping or Filegroup Stripping