There is alot of difference between having 100 5MB documents and 5M 100B documents.
If your test doesn't give performance differences, look at
other parameters:
Using the file system is invaluable during
testing, you can just look at your files.
OTOH, backing up millions of little files is a pain compared to backing up one database file.
One more question: Do you need any kind of concurrency?
This could also influence your decision one way or the other.
In reply to Re: BerkeleyDB vs. Linux file system
by Jost
in thread BerkeleyDB vs. Linux file system
by perrin
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |