[nycbug-talk] Fwd: Re: OpenSolaris dead - storage benchmarks

Fri Aug 27 18:03:53 EDT 2010

On Tue, Aug 17, 2010 at 12:40 AM, Charles Sprickman <spork at bway.net> wrote:
> On Mon, 16 Aug 2010, Jason Dixon wrote:
>
>> (Sorry for the ill-formatted version, resending)
>>
>>> On Mon, Aug 16, 2010 at 9:48 PM, George Rosamond
>>> <george at ceetonetechnology.com <mailto:george at ceetonetechnology.com>>
>>
>> wrote:
>>>
>>>     I have not read the whole document, but certainly there's open
>>>     questions about ZFS (which I have heard is a full, irrevocable
>>>     fork), dtrace, etc.
>>>
>>> If its a full fork, and non Sun/Oracle employees have been porting it to
>>> FreeBSD, whats the big deal there?
>>>
>>>     I'd assume that there's a wide audience of developers and users who
>>>     will be looking for a new home. . .
>>>
>>> Solaris was only open source for a few years. Other than those that
>>> jumped on the Solaris band wagon after the open sourcing, I'm not sure
>>> who feels they are in need of a new home. Also, thats only a big deal
>>> for kernel and some core userland developers.
>>
>> You're vastly underestimating the number of Solaris/OpenSolaris users and
>> companies looking for an exit path.  They were enticed into OpenSolaris
>> and SXCE thanks to exciting projects like Crossbow and bleeding-edge
>> features in ZFS.  Now many of their systems are backwards-compatible with
>> anything that Oracle will support now or in the future.
>
> I've noticed that the Linux folks are also quite excited about this - it's
> an opportunity to push btrfs pretty hard.  I do think many of them are
> overestimating how close to production use it is though.
>
> There is also quite a bit of talk on more linux-centric boards about zfs
> being a poor performer compared to just about any linux fx paired with
> lvm/md.  This I'm not so sure about - Phoronix seems to be the go-to place
> now for benchmarks, and some of their methods are... odd to say the least.
>
> Slightly OT, I'm looking for a benchmark that deals more with iops rather
> than throughput and has a fairly simple output - just something that sweeps
> through various sizes and then spits out a summary of how many io
> operations/second it recorded.
>

bonnie++, i've used this one
iozone, i haven't used it, but it looks good

>>> Don't get me wrong, things are getting interesting from here, but I
>>> don't think its going to be that big of a deal in retrospect. The only
>>> big deal outcome is if Oracle ends up re open sourcing it. If Oracle
>>> increases Solaris's popularity after close sourcing it, it proves Oracle
>>> is a better business than Sun, not that Close source is a better
>>> business model than Open Source. You have Linux and the BSDs proving the
>>> viability of Open Source OS development.
>>
>> This is not a closed-versus-open source issue.  For OpenSolaris users,
>> it's having a predictable support path and roadmap.  For the BSD
>> community (in particular, FreeBSD), it's an opportunity to a) embrace new
>> users and b) show off the capabilities and innovation available within
>> BSD.
>>
>> I've had countless discussions with other Solaris users over the last few
>> months.  FreeBSD is a popular topic of discussion, particularly as a
>> favored alternative to Linux.  But there are still a lot of questions
>> these folks have, and I think this is a good time to reach out and
>> demonstrate why BSD software (and the associated communities) are a great
>> option to wayward Solaris users.
>
> I am somewhat concerned about FreeBSD in general and specifically what the
> plans are there for ZFS (both in general and  the impact of oracle perhaps
> not ever releasing any additional code).  I've seen lots of folks
> experimenting with FreeBSD just for ZFS, many that have never given FreeBSD
> a second thought.  They seem to avoid the official mailing lists and work
> with each other to tune and debug things either on the FBSD forums or other
> general tech forums, so I don't know how aware the core group of FBSD
> developers are of this growing group of users coming to FBSD strictly for
> ZFS (and finding a pretty decent OS to go along with it).
>
> It seems like even with ZFS now being offically labelled as production-ready
> on FBSD, there's very little documentation popping up, and the few people
> that understand the internals are not really available to help out those who
> want to document it.  The classic case is the whole issue of running ZFS on
> i386 - tuning it to not panic is not easy, and the manpages are still just
> straight rips from open solaris - there's no FreeBSD-specific info in there.
>  The wiki (http://wiki.freebsd.org/ZFS) is somewhat dated and does not
> really explain the i386 tuning issue well. It's also not open to editing.
>  There are a few tricks to getting i386 stable, but it's not readily
> apparent.  The biggest hint I found (and I've heard this may have changed)
> is that the arc_max value is NOT a hard limit, but a high water mark where a
> thread to flush the arc to disk kicks off.  On slower/older hardware with
> unpredictable disk load, the time it takes for that thread to do what it
> needs to can fall behind the rate at which some process is writing data that
> lands in the ARC.  When that happens, boom!  I had much better luck tuning
> this once I realized that I could routinely load the box in a way that the
> arc could grow to double it's "max" size.  In some cases that meant just
> going with better hardware (and 64-bit, where zfs is very stable), and in
> others it could be worked around by setting the ARC max very low - in
> essence giving the flush to disk thread a little bit of a head start.
>
> There's also the issue of sysinstall, especially if you want to do a zfs on
> root setup.
>
> And there's currently a show-stopper bug in 8.1 where a raidz1 array that's
> degraded cannot be booted from because some regression was introduced into
> the zfs bootloader.
>
> Anyhow, yeah this could be a great opportunity, but I think there has to be
> some sort of meeting of whoever is "in charge" of the FreeBSD project to get
> some kind of leadership/management going on this front.  Docs would actually
> be an excellent start, but beyond that I think there needs to be more
> ownership of ZFS amongst the development team - right now I think there's
> essentially two people involved in what could be one of the biggest features
> FreeBSD has ever introduced.
>
> BTW, if anyone wants to start a general zfs thread, I love to talk about it.
> :)  I'm in the midst of migrating a bunch of 4.11 boxes to 8.1 w/zfs on
> root.
>
> Charles
>
>
>> --
>> Jason Dixon
>> DixonGroup Consulting
>> http://www.dixongroup.net
>> _______________________________________________
>> talk mailing list
>> talk at lists.nycbug.org
>> http://lists.nycbug.org/mailman/listinfo/talk
>>
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org
> http://lists.nycbug.org/mailman/listinfo/talk
>


-- 
-jesse