Pro tip: Defragmenting only works on spinning drives because it puts the data nearer to the spindle so seek times are shorter. Solid-state drives wear out faster if you defragment them, since every write involves a little bit of damage.

[–] [email protected] 6 points 4 months ago* (last edited 4 months ago) (1 children)

Random reads are still slower than sequential in SSD. try torrenting for a year on SSD, then benchmark then defragment then benchmark. it will be very measureable difference. you may need some linux filesystem like XFS as im not sure if there is a way to defrag SSDs in windows.

[–] [email protected] 4 points 4 months ago (1 children)

That's because the drive was written to its limits; the defrag runs a TRIM command that safely releases and resets empty sectors. Random reads and sequential reads /on clean drives that are regularly TRIMmed/ are within random variance of each other.

Source: ran large scale data collection for a data centre when SSDs were relatively new to the company so focused a lot on it, plus lots of data from various sectors since.

[–] [email protected] 1 points 4 months ago* (last edited 4 months ago) (1 children)

I'm pretty sure running XFS defrag will defrag without trimming no matter the type of block device.

Edit: yea you might actually be right. I Played with my fstab too much years ago, and never thought of that untill now

[–] [email protected] 2 points 4 months ago (1 children)

I understood that XFS automatically mounted SSD's with XFS_XFLAG_NODEFRAG set? Is this not the case?

[–] [email protected] 2 points 4 months ago (1 children)

yea you might actually be right. I Played with my fstab too much years ago, and never thought of that until now

But does that flag affect manually running xfs_fsr?

[–] [email protected] 2 points 4 months ago

According to the man(8) page, it will avoid touching any blocks that have the chattr -f flag set, which is XSR_XFLAGS_NODEFRAG... So I think if the docs are still accurate to the code, yes.

A lot of ifs in that assumption.

load more comments (22 replies)