[RESOLVED] Which one is faster? file_exists() or query the table

niroshan

Hi Guys,
I have an application where i enable users to upload files. While uploading i rename them into my own naming convention and save them in the server.

Before saving it in to the server I always check whether that file name already exists or not using file_exists().

This application has large user base so daily uploads most of the times comes to over 1000 files. each file upload insert a row into the table with relevant file name.

So i`m just wondering which method is much efficient?

Query the database and check file name exists?
or
use file_exists function directly.

Thanks in advance.

Best regards,
Niroshan

johanafm

Microtime both and compare. Also, it wouldn't surprise me if file_exists is faster when you have two files in a directory than when you have 1 million files in a directory, so it could be that file_exists is faster than DB for less than N files in directory, and DB being faster for more than N files in directory for some N, which also might differ depending on both file system and db index type and index length.
Also, if you have a constantly growing amount of files, you really should split them up over several subdirectories, possibly more than one level deep. While the issues you may encounter are likely to differ between file systems, you should probably not go above 100k files per directory.

NogDog

Just a thought: if the file name column in the DB table has a unique index, then you can just go ahead and do the insert, and as long as everything is copacetic it will run and return a valid resource. If the file name already exists in the DB, then the query will return false and you can check the mysql_errno() (assuming the MySQL DB extension is being used) to see if it's a duplicate key error (1062).

That way there's no extra processing unless/until you hit that error, which assuming your file naming scheme is highly unlikely to generate duplicates (e.g uses uniqid()), should be a very rare occurence.

niroshan

Thanks johanafm,
I will try the benchmarking...
In my case each file represent a single row in the table will result in both the file count and row count will be same. So I guess its better to query the table instead of search the file system since the N no of files expected to grow large.

Thanks NogDog,
I think i did a mistake of not telling you guys the naming convention i used in the first place. AS a application requirement I have to preserve the original file name which uses used so what i do is, I check the file name using file_exists() and is exists i concatenate the filename with the current time stamp.
ex:

$newFileName = <User file name>.'_'.date('Ymd').'.flv'; //assuming its a .flv file

I does it inside a while loop until it return false.

But i guess as you suggested concatenating uniqid() is much better approach then using time stamp since this will reduce the no of rounds in the while loop. Thanks for the advice.

Thanks all of you for the advices .. really appreciate it.

Thanks,
best regards,
Niroshan