I can think of a lot of innovations,
what I really want are features like
- simple clustering on a service level and additionally storage level to
improve availability of NFS/SMB or optionally iSCSI (not as complicated
or feature rich as RSF-1 or the Linux Cluster Filesystems like Ceph) to
allow a failure of a storage server AND a server that offers services.
I have this working for a master/slave server pair using ZFS itself as
Cluster Manager for SMB and NFS as in Solarish an export + import on
another head restores the services automatically. iSCSI requires some
more work as this is no longer a ZFS filesystem property. Solaris is
perfect for this simple scenario as NFS and SMB are fully integrated in
ZFS. I use two iSCSI targets on two storage server created from a whole
ZFS pool that I mirror on the master one via an Initiator. The needed
failover between heads is scripted with a remore control as well as the
failover of services and targets via virtual ip adresses. Very simple
but it works well. Only problem is performance (iSCSI over ZFS targets
on a ZFS pool) and a weird behaviour of ZFS and the initiator. If your
pool is managed like a single disk and ZFS/the initiator looses
connectivity, you get an io error that you can only fix with a reboot. A
pool export with missing disks is not possible - really a problem not
only in this Cluster Case.
Would allowing a pool export/destroy when all disks are missing a big
problem?
There was AoE from Coraid (ATA over Ethernet) that may be suited or
Nexenta's continous replication as other options to improve
availability. A Cluster fiilesystem ontop of ZFS may be another option
but I prefer a simple solution based on ZFS + SMB + NFS (iSCSI) only
Encryption
There is no suitable encryption outside Solaris. Datto/ ZoL is working
on it, a very welcome improvement. An innovation may be not only
filesystem encryption but optionally user file content optionally
filename encryption on SMB based on user or SMB group (sid) login
credidentials so every user or group has his encrypted data on a common
share based on his name + pw.
SMB3
should be upstreamed from NexnetaStor
Simple Restore from the iso installer menu (restore from datapool)
with a backup command to save the whole rpool from a running system to a
datapool.
Gea
@napp-it.org
Am 14.09.2017 um 15:06 schrieb Peter Tribble:
On Wed, Sep 13, 2017 at 9:14 PM, Miles Fidelman
<[email protected] <mailto:[email protected]>> wrote:
There does seem to be some work on micro-kernals, environments
that run on bare iron (e.g., Erlang on bare iron has potential),
Yes, I've spent time looking at unikernels (and rump kernels),
although they're
really fat processes that still need an OS layer above to coordinate them.
and continuing work on cloud stacks. But still... not as much
innovation as one might hope for. And my personal bugaboo - work
on single-system-image clusters has basically disappeared.
That's a huge gap. But most people throw an ETOOHARD error and rebuild
everything in microservices and deploy an army of squirrels.
Whether you have to do the single image all through the OS is another
matter,
though. You can imagine a shared filesystem (and that's another area
needing
more work) coupled with something like distributed SMF doing essentially
the same thing.
Miles FIdelman
On 9/13/17 1:01 PM, Peter Tribble wrote:
In my OpenSolaris t-shirt collection, I have one with the slogan:
"Innovation happens everywhere"
I'm not sure this is *entirely* true; Solaris 10 was a massive nexus
of innovation that has proliferated out to other operating systems
over the last decade. Frankly, there's not much else been happening
in systems development.
From what I can see, between the cloying boredom of Linux monoculture
and the dead hand of POSIX "standardisation", systems have stagnated.
Even in illumos, we're largely doing a bit of light gardening - a bit
of weeding here, a bit of pruning there, replanting the odd bush. But
no real landscaping is being done.
Which begs the question - is systems innovation done and dusted?
Or is there more to come?
And if there is more, what sort of new features are wanted?
At which point I open up the floor to anyone who wants to contribute.
(Note: I'm not talking about a gaps analysis. We [illumos] need more
drivers, more applications ported. We already know that, and it's
just
copying, not innovation. So there is an interesting subject
there, but
if someone wants to follow that then please create a new thread.)
Cheers,
--
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
--
In theory, there is no difference between theory and practice.
In practice, there is. .... Yogi Berra
--
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
*illumos-discuss* | Archives
<https://illumos.topicbox.com/groups/discuss/discussions/T83f198c8597cf8e3-M7630634b12ff1dc4069dfa36>
| Powered by Topicbox <https://topicbox.com>
--
------------------------------------------
illumos-discuss
Archives:
https://illumos.topicbox.com/groups/discuss/discussions/T83f198c8597cf8e3-Mbc53a3a744b8f454f7194451
Powered by Topicbox: https://topicbox.com
