Thanks for the detailed information Al.

>>Inline:
>>
>>
>>>I'd leave it to the OS.
>>
>>OK, but you do mention using the Windows defragger under a particualr circumstance. My question is then; if a defragmentation is needed, would you recommend the Windows supplied defragger or a commercial defragger (who claim to be superior in various respects).
>
>The circumstance is if a volume is, or was, completely full. In that case, after freeing up space defrag is usually required to restore optimal performance. By default, Windows built-in scheduled defrag runs at low priority, and if the server/volume is busy most of the time, scheduled defrag may take a long time to complete. That's why I'd recommend a manual defrag, one time, in that particular circumstance.
>
>NTFS is not immune to fragmentation, but it is resistant to it unless a volume gets very full.
>
>IMO a "happy" server is one whose disk activity light is not constantly lit, maybe half the time or less. In that case you are not likely to see any improvement in real-world performance between the built-in Windows defragger, and a third-party product. OTOH if the array gets hammered for extended periods, maybe you can use any performance edge you can get, however slight it may be. YMMV.
>
>>AND - would you run the defragger on a schedule (as Windows recommends when you open the Windows defragger and which William's post implies mu st be done since otherwise how would Windows be handling the defragmentation issue itself) ?
>
>Yes.
>
>>AND - is there any danger in running these things with a RAID setup?
>
>Defrag makes sense only for single ( mechanical ) drives or simple RAID types ( 0, 1, 5 etc. ) that don't include advanced controllers. Generally, defrag utilities use transaction-type processing, making certain a cluster has been successfully copied to a new location before deleting the original, with mechanisms to help avoid problems in the event of power loss, hardware failure etc. While what a defragger does can sound scary, in practice it should be as safe, or safer, than any regular file copy or move operation.
>
>Trying to defrag SSDs is harmful, as it increases wear for no gain. I believe Win8 and Server 2012 detect SSDs and will not let you try to defrag them; I think in some cases you can TRIM them instead.
>
>High-end drive arrays and SANs typically have their own advanced controllers. They logically remap I/O so the simple cylinder/sector/head reorganization done by a defragger has no meaning to them. You generally treat them as "black boxes" that handle their own performance optimizations.
>
>So, where it makes sense, defrag is not dangerous. It's been included in Windows for a long time; if it were dangerous the Interwebs would be full of horror stories. It just works.