• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

RaidZ2 Storage using SSD's?

A

Apollo686

Limp Gawd
Joined
Nov 15, 2004
Messages
227
I'm creating an array for ESXi storage and planning on using SSD's. I know they can be used for L2ARC and ZIL, but I'm wondering if anyone has experience using an all SSD array. I'm planning to run them off LSI 9211's (flashed with IT FW) and just wondering if there are any configuration issues or performance tips I should be aware of. Planning on running 6x SSD in RaidZ2.
 
I have found references that claim that ZFS has supported TRIM for awhile now. You absolutely will need TRIM for a ZFS vdev, since all writes except for uberblocks are COW, which means successive writes are always to a different location, so fairly quickly the disks will fill up, from the TRIM point of view.
 
Thanks for the tip. I'm using Solaris 11; I will have to look into TRIM support for vdev's unless anyone knows more about this.
 
Also, it's worth noting that if a HW Raid 5 on an LSI 9260 will yield better performance or fewer problems I will go with that rather than RaidZ.

Thoughts?
 
If you are considering ZFS vs HW Raid, I would recommend the former. The performance should be no worse, and you have the various ZFS features available (assuming you care about those - not everyone does...)
 
Hardware RAID doesn't have trim yet, so there is that. And I'm not an expert, but wouldn't building an array fill the SSDs to the brim ?
 
Wouldn't it not matter if it had trim or not since writes happen every 5 seconds? (ie. enough time between writes for the drive to figure itself out?)
 
How would the drive figure out itself? Remember it will not have a builtin understanding of anything zfs like it does for ntfs.
 
Isn't the action for non-TRIM usage, when a write is initiated to first blank the memory cell, then to write data to the cell? (in other words 2 writes per operation)

Whereas TRIM allows the OS to actively blank cells during idle time to prepare them for write?

So I would think operation would still be pretty spectacular even with the performance hit?
 
drescherjm : I haven't seen any proof that there was an SSD capable of understanding NTFS. I'm not sure that would be a good idea anyway, unless it was 200% bulletproof.

staticlag : what you describe should only happen during very heavy usage. For a smaller quantity of writes, there should always be blank cells to write to : in the reserved area.
 
You must log in or register to reply here.
Back
Top