1
1
I am experimenting with multiple cluster filesystems for serving websites, and GlusterFS seems to be closest to what I would want to have, both feature and stability-wise.
However, when issuing a find
command on a large number of files, I am experiencing a considerable slowdown compared to native disks. (About by a factor of 3.)
GlusterFS, according to the description, does have some sort of local metadata cache, so in theory the find operation should be reasonably fast. I have also noticed that sometimes a simple ls
command on a folder with ~200 files takes multiple seconds to complete.
Question: How can I nail down the cause of this performance issue, and can this kind of lookup be made faster with GlusterFS? Is this a possibly just down to FUSE being slow?
Setup:
- OS: Ubuntu 16.04
- GlusterFS version: 3.7 and 3.8