File Search In Windows 7 Is Unreliable?

It seems that the search functionality in Windows 7 cannot be relied upon. For example, I have a folder with around 2000 files. Search indexing has been enabled for this folder for about 12 months, yet when I try a simple filename search it fails to locate files that I know exist. In explorer, I click in the search box and type "glebe" and hit enter. I know that there are seven filenames beginning with "glebe", plus two other filenames that contain the word "glebe". All files are more than six months old, and thus have had plenty of time to be indexed. It only finds two of the seven filenames beginning with "glebe", and none of the filenames containing "glebe". There does not appear to be any special feature of those two files that differentiates them from the others.

In addition, if I created a file a few hours ago and then try to locate it, there does not appear to be any guarantee that the file will be found. Thus the search is of no use at all; if I can't be certain that every file is found every time I search, I can't be certain that I'm not corrupting my data when I manipulate the files that the search does find because I may be leaving files unedited that must be edited. As the delay between a file being created, renamed or deleted and that change being indexed cannot be predicted by the user, even if the index did work correctly, I cannot be sure of how long to wait before the results become reliable. Therefore, even if the search system could be relied upon to deliver the correct results, this undeterminable delay renders it unusable.

I have tried reindexing the folder, which does not fix the problem but does provide a set of search results that are different, yet equally unreliable. I have tried this on several hundred computers running Windows 7 with identical resultsdditionally, the indexing service causes my computer to grind the hard disk for a large periods at random intervals, making my work inefficient. If I want to search for a file I am happy for the hard disk to be accessed to the extent that the file is located, but I don't want my ability to work efficiently to be impeded by indexing millions of files that I'm statistically unlikely to ever search for.