A. Consider that a RAID 1 set is two disks, with each storing the same information. You have one mirror set for the database (ntds.dit and SYSVOL). You also have one mirror set for the logs, giving you four disks total. You also normally have another, separate mirror set for the OS.
In most environments, the Active Directory database volume performs a huge amount of I/O while the log volume performs almost no I/O, which means you're wasting two disks worth of /IO. Most experts today agree that instead of two RAID 1 configurations, you get far better I/O and performance by combining the four disks suggested above into a RAID 5 or RAID 10 configuration. Under these RAID levels, all of the disks can be used primarily for AD database purposes, instead of half the disks sitting mostly idle for logs. Because you have RAID, even if a disk fails all your data is still available, so the concerns over placing the database and logs on the same disk are minimized. (But not totally written off, because a logical NTFS corruption could still cause all data to be lost. This is very unlikely compared to a disk failure, however.)
The only time it becomes more important to separate the database and logs is if you have non-RAID disks, in which case you should separate the database and the log files onto separate drives.Related Reading:
- Q: I want to install an enterprise certification authority (CA) on one of my Active Directory (AD) domain controllers (DCs), but it only lets me install a Standalone CA. Why?
- Protect AD from Administrative Errors
- Q. I read on MSDN that you can only create a forest trust from a Windows Server 2008 forest to another Server 2008 forest. Is that correct?
- Q. What happens when a read-only domain controller (RODC) that's a DNS server receives a DNS write request from a client?
Check out hundreds more useful Q&As like this in John Savill's FAQ for Windows. Also, watch instructional videos made by John at ITTV.net.