ZFS Mysql / MariaDB server on ZFS - better 2x800GB enterprise SSD or 8x300GB 15k rotating rust?

R

rainer_d

Hi,

I seem to remember postings about performance being less than optimal due to copy-on-write?

I need to replace a HP DL380Gen7 server with a Gen10.

Back when these were procured, HP simply didn't have any reasonably priced SSDs.
They still don't, but due to market-pressure, we get good discounts.

Anyway, back then we used two 600GB disks as boot-volume and 6x600GB for the database (and as a file-server, before it got too big and we had to have a dedicated server just for files)
The pool just has 50GB allocated.

As such, I would have said 2x800 GB "mixed use" SAS SSDs in RAIDz1, underprovisioned to 400GB each would deliver enough IOPs and provide enough reliability?

Can anyone relate any real-world data?
 
Do you mean a mirror or actually Raid-Z1?

I would expect the much shorter access times of the SSDs to provide a noticeable difference.

I set up a server the other day with a ZFS pool for boot/storage, but used a UFS gmirror of 2 SSDs for MySQL. Things are probably better now but I'm still not convinced that ZFS is an ideal file system for databases.
 
Ah, sorry. I mixed it up.
ZFS mirror it is. I need to use ZFS because I need to be able to send ZFS snapshots to MySQL slave servers.
 
You must log in or register to reply here.
Back
Top