Re: graid hit the tree

Alex Dupre writes:
| Alexander Motin ha scritto:
| > You can choose. Their functionality is comparable, but with graid:
| ...
| 
| Yes, as I supposed, so the answer is 'yes'. Obviously it needs a lot of 
| testing. Thanks for your work.

Yes, it can use more testing which is why we (Cisco) wanted it in the 
tree.  However, our experience has been very positive in performance
and reliability.  It has passed some of our harder failure tests of 
which others fail (race conditions of changing data while a rebuild is
happening etc.).

Now I don't have to maintain our hacked together ata and ata-raid
changes :-)  We are very happy to see this get into mainstream 
FreeBSD so others can help find and fix problems.  It's very nice 
that it can work on any type of drives and not be tied to ata disks.
This means it can/could work with scsi controllers with sw raid
bios extensions etc.  We have been testing it with ahci and and
cam attached ata disks.  If the hw supports it, it can update
status leds.  ata/ata-raid has served us well but graid will take
us into the future!  It was designed from the start as something
that could be extended and deal with various failure modes.

A big thanks to Alexander, Warner, iXsystems and Cisco for making
this happen.  It was Scott's vision a few years ago and now it is
a reality.

Doug A.