[zfs-discuss] Getting a handle on the current weak areas of ZFS-on-Linux

Uncle Stoatwarbler stoatwblr at gmail.com
Thu Aug 22 19:50:32 EDT 2013


On 20/08/13 23:49, Anthony D'Atri wrote:

> I looked at Nexsan's E60/E48, has some nice features like environmental/status info that can be slurped off the box for automated monitoring.  But the largest volume it can present is only 64TB, and one can't have multiple parity groups per volume.  It was like going back to 1995 and Oracle's ZFS storage appliances actually weren't completely out of the pricing ballpark.

Nexsan have -E versions of the devices (E60-E and E48-E) which are 
simple JBOD drawers. The problem is that their ZFS head units are sold 
attached to E60/E48 Raid systems and only the bottom end are "proper" ZFS.

None of their ZFS systems have enough ram to serve more than ~40TB in 
any case.


>> As the appliances are all far more expensive than a Nexenta-based solution at 16TiB, that paths makes the most sense for that size and TrueNAS makes more sense at large scales (I'm assuming that the OP wants commercial support).
>
> At some point does Solaris 10/11 make sense?

Oracle quoted us $stupid_money. They're clearly not interested in large 
storage systems anymore (dumping the thumpers was a pretty good 
semaphore for that)

On top of that there's strong resistance here to learning "yet another 
OS" and some sour tastes about past experience with Sun product (I 
managed to miss all that, as it was before I started working at my 
current location)


To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.



More information about the zfs-discuss mailing list