Puzzle of disk analyze speed

[Aimar]

MR Kelderek:
 I have read the source code of your disk defragmentor,it's very great.
it defrags rapidly..But if i want just only to write a disk analyze function,
according to your method,it seems to slow.
 But i deep into the Diskkeeper and the utility of windows,they both analyze rapidly,is there any APIS undocumented? Or your method isn't fit for the just analyze?

[jeroen]

Well, if there is an undocumented function in windows and anyone can tell me about it then I'd love to hear about it! At the moment I am forced to walk through all the files on disk with FindFirstFile() and then call FSCTL_GET_RETRIEVAL_POINTERS for each and every file to discover where it is physically located and wether it is fragmented. Which is very slow. By the way, most of the time is spent in the FindFirstFile() function.

See the Microsoft website for more information.

[Aimar]

Well,I have test some defragmentors on FAT32 filesystem,i find that no defragmentor can analyze rapidly like on NTFS.So,i doubt that the NTFS filesystem must record some fragment info secretly.what about your opinion?

[jeroen]

I suppose it's possible that NTFS records secret information about fragmentation, but I think the Microsoft defragger has faster analyzation because it uses undocumented (read "secret") system calls. It's a well-known fact that there are a lot of undocumented system calls in Windows, perhaps Microsoft keeps them secret to have an edge on the competition.

[jeroen]

By the way, my defragger does not have a separate analysis mode. While analyzing it does a lot of things, such as saving information in memory and updating the screen. A specialized analysis would be faster. It could skip all files that are smaller than 1 cluster, for example, because they cannot be fragmented. I don't think it would be as fast as the special analysis function in the Windows defragger, though.