Creating Multi-level directory structures dynamically (theory help)

[Cent]

Hi All,

Been trying to figure out a solution to begin coding an image hosting script. The don't want to store images into the MySQL database so I'm looking at a decent filing system. I've seen a few image hosting apps which breaks down the user's ID # to formulate a directory structure for any images uploaded. I've also seen ways of using a person's username like p/e/t/peterpan_image.jpg.

Is that the best way to organize what could potentially be thousands if not tens of thousands of media files? Would it be better to use the username or ID # concept? How many levels should I go down?

Which method do you think would work best? Does it matter?

1. User ID # 1024..... 1/0/2/1024_img.jpg
2. Username PeterPan... p/e/t/peterpan_img.jpg

Each image will have a unique ID.

Any thoughts or past experiences are gladly appreciated. If anyone knows of any open source scripts that use either method, any URLs would be great.

Best,
Cent.

[jazz_snob]

Cent wrote:

Is that the best way to organize what could potentially be thousands if not tens of thousands of media files?

Why not? I have programs that store user data by user name and it works quite well, provided each file name is unique.

1. User ID # 1024..... 1/0/2/1024_img.jpg
2. Username PeterPan... p/e/t/peterpan_img.jpg

I don't quite see how thats of any benefit or how it even works? Fill me in. If user id is #1024 why not:

/path/to/the/data/1024/someimage.img

[Cent]

hi jazz_snob,

I had thought of just using: /path/to/the/data/1024/someimage.img

but that brought up another question which I couldn't find an exact answer to... Is there a maximum number of directories that can be within <data>? Once you get past a few hundred or perhaps a few thousand most filesystems hit their limit.

Right now I’m working on a *nix box so it appears the number varies based on the kernel and any patches installed. I’ve read varying reports from 15k to 100k for files (since directories are considered files as well) in a single directory. So I was looking at ways to increase that number by creating subdirectories 1-3 levels deep.

[jazz_snob]

This is a great question. And the max # files would obviously depend on the OS, file system and configuration.

If found this which may be of interest here:

The ext2 inode specification allows for over 100 trillion files to reside in a
single directory, however because of the current linked-list directory
implementation, only about 10-15 thousand files can realistically be
stored in a single directory.

I think more googling is required.

[madwormer2]

Lol, funny story, you know how error logs are created by some programs, such as corporate firewalls and stuff, well, I was working at BT for my work experience, and the person I was working with crashed a computer just by opening the log file folder.

Turned out there was about 12,000 1kb files just sitting there, and they wouldn't be listed (windows, need I say more.) We had to delete them via the command line ^_.