[rescue] ZFS Merkle trees/git graph integrity - Re: Has freshmeat.net disappeared forever?

PAK pakenned-list at pobox.com
Wed Jun 25 09:39:09 CDT 2014


On Tue, Jun 24, 2014 at 07:31:45PM -0400, Toby Thain wrote:
> On 24/06/14 6:08 PM, PAK wrote:
> > [...] git history is essentially a cryptographic composed graph
> 
> FWIW, ZFS has the same property and the same habit of exposing
> hardware faults that are hidden by filesystems that don't have
> integrity checks.

It's funny--I read about this sort of cryptographic composed graph in a
research paper (back in grad school) and I _wanted_ it then.  I noted
how soon I had Sun engineers touting the feature (and trying to sell us
on it).

I no longer manage large systems or large deployments, so my
sysadminning is personal (and not at the scale of some of the personal
collections on this list--oh my).

There have always been constraints that have limited my (personal)
opportunity to deploy a system supporting ZFS.  Right now it's space and
decibel limits.  So I try to content myself with using git where it's
important on top of a filesystem on top of LSI RAID card.

*sigh*

But yeah, I would have ZFS if I could :)

Cheers,
PAK


More information about the rescue mailing list