I have a process where I am removing BLOBs from a DB and writing them to a file structure on disk. I have a 1TB disk, with a single folder in the root, and a structure similar to:
There will be potentially several thousand 'file.txt's in the same folder, but the average appears to be about 300. I ran my export program for 10,000 records to gather metrics, and now have concerns.
Each file.txt is compressed with
DeflateStream, and appears to average about 900 bytes. The disk was formatted NTFS with 2k clusters, and 8dot3 filenames disabled. Though, based on the attached screenshot, each file is taking 8192 bytes instead of 2048 like I would expect. I have over 75 million files to write, and we add about 3 million a week, so disk space usage is something that needs to be addressed.
So, my question:
Why is each file taking 8kb even though I used 2kb clusters when formatting?