“There has been work by the Lawrence Livermore National Laboratories in porting ZFS to Linux as a native Linux kernel module. This LLNL ZFS work though is incomplete but still progressing due to a US Department of Energy contract. It is though via this work that developers in India at KQ Infotech have made working a Linux kernel module for ZFS. In this article are some new details on KQ Infotech’s ZFS kernel module and our results from testing out the ZFS file-system on Linux.”
I don’t like Phoronix. So often, it seems that they don’t understand what they are benchmarking. I’d imagine the discussion to benchmark ZFS went something like this.
“Hey. I’ve got an idea. I just got my hands on a pre-release ZFS Linux module that nobody else has access to. Let’s benchmark it!”
“Well, we don’t have a test system that’s at all representative of what people who need ZFS would be using. I mean, it’s only got half the ram that an actual production system might be using at an absolute minimum. Plus, our benchmarks don’t run continuously, so the heavy amount of caching that ZFS does won’t be utilized, skewing results in a dis-favorable manner. Never mind that this is early code that hasn’t even been released yet.”
“Yeah, but, people won’t realize that until they click through a large amount of pages, each crammed full of as much advertising as possible.”
…
Why did they only include Solaris on a couple of tests? They don’t say.
“we don’t have a test system that’s at all representative of what people who need ZFS would be using.”
That won’t stop some people from using ZFS, anyway. They’ll use it simply because they can.
Still, the benchmarks are useless. They even pass judgement on ZFS using them.
It’s like saying Winodws 7 isn’t ready, because it runs like crap on my system that has only 512MB of RAM.
And you will never see it shipped with linux. Patents covering ZFS are still being argued over in court. NetApp and Sun never settled. Who knows what Oracle will do with the technology. There is no way Linus will ever let something possibly tainted by patent issues ship with the kernel.
Shipping with the kernel has nothing to do with patents in this case. GPL & CDDL incompatibility prevents it in the first place.
NetApp and Oracle have agreed to dismiss their lawsuits.
http://www.netapp.com/us/company/news/news-rel-20100909-oracle-sett…
From the Phoronix article:
One of the main features of ZFS is storage pooling, and getting rid of all the hassles of managing multiple disks. Testing it on a single disk without testing it on multiple disks is pretty much pointless.
Sure, some people run it on their laptops with a single disk, to get access to all the snapshots and checksumming. But that’s an edge use-case, not the primary one.
Sometimes, I think Phoronix just “benchmarks” random things, throws them up on the ‘net, and waits to see what sticks. There’s really no statistical methodology in place.
On the flip-side, the race is now on to see who gets ZFSv28 first: FreeBSD 9.0 or Ubuntu 10.04 LTS (and whatever other distros KQI supports)
There are ZFSv28 patches available for 9-CURRENT right now, but nothing committed to the source tree as yet. And the Linux kernel is in “limited beta”. Both with a release date of “early 2011”.
and FreeBSD 9 should be out mid summer