Main | Subversion: Merge trunk modifications into a development branch »

October 30, 2009



Its a nice article indeed.

Michael Andrews

I'm glad you enjoyed it. Always good to get feedback. ~Michael

Blogs are so informative where we get lots of information on any topic. Nice job keep it up!!


Thank you Michael.

Your article is very balanced between theory and coding. I have to generated 120K file every month and didn't know how to store them efficiently. It is done now.

Michael Andrews

@Ludovic - Glad to help!


i don't understand the DirectoryHashTest class' can you write it in simpler way?

Rich Sadowsky

really great explanation. I've been using this pattern for nearly two decades for repositories with large number of files. I was recently struggling with explaining the purpose of this pattern to one of my fellow engineers. I googled "file directory structure hash filenames" and this was the first result. Bravo.

BTW, I've used this with billions of files in production systems. It really does work.

Michael Andrews

Rich - Thanks for your post! I enjoyed writing this one. So many people implement complicated directory structures for hashing - with dates, and timestamps, and user names, etc - find this simple approach fits most of the use cases. Thanks again! ~Michael


I have a problem with hash code in our software (I am buy this software). From your post, I can figure how this software create a hash code. Thank you for your tutorial :)


I spent a few days trying to create an optimal directory structure for my RSS directory site to store XML feeds. Thanks for writing this. I am now at peace :-)

Autymn D. C.

Write one for a Base64 path.


how about in C language ?


Interesting. I am planning to add this hash logic to my application. I needed to do a hash with filename and size mainly.


Clever boy!


Your fileName is probably uniformly distributed to begin with, hence you have really proven nothing in the last part of the article.

The comments to this entry are closed.