Media response to ZFS

a real threat to companieslike EMC, a triumph for open source, a significant, long term, Unix advantage over Windows, another key milestoneachieved in the move to network computing, and arevolution in storage technology

Two weeks ago I blogged about the media's unthinking response to an uninformed and ungrammatical Sun press release announcing the Niagara (or UltraSPARC T1) CPU.

Sometime later that drew a response from a senior non Sun analyst I frequently exchange email with. Here's part of what he said:


Perhaps part of the problem is often you have to describe what something *isn't* as well as how it is better.

Eg: Niagara - it is NOT yet another CPU designed as a "jack of all trades, master of none". Rather, it was built from the ground-up to be as naturally efficient at multi-threaded tasks as possible.

That might excuse media ignorance but of course says nothing about the problems at Sun's "outbound marketing" operation.

Last week, however, ZFS, the zetabyte file system announced with Solaris 10 last year, finally made it into early release. This got a header and two paragraphs in an omnibus press release headlined "Sun Announces Support for Postgres Database on Solaris 10."


Solaris ZFS - The Most Advanced File Systems on the Planet

Sun today released its next generation file system, Solaris ZFS, to the OpenSolaris community. The ground-breaking capabilities of Solaris ZFS include self-healing data through advanced error detection and correction, task automation that dramatically simplifies storage management - in some cases reducing task times from hours to seconds - and built-in storage virtualization that eliminates the complexity of a volume manager. ZFS will be broadly available to customers in the first half of 2006.

"Solaris 10 provides material enhancements from prior releases. It highlights Sun's commitment to continuous innovation and puts Solaris ahead of the competition," said Michael Minichino, IT director at Parago. "Incorporating ZFS into the Solaris Operating Environment presents a compelling value proposition as it will compete with other solutions cost burdened with expensive software such as Veritas Volume Manager and File System at a much lower price point. This combined with Sun's new server products will put a high end UNIX(R) solution within reach of cost conscious small and medium enterprises."

Since ZFS is probably the most important user accesible change in Solaris 10, you'd think burying the announcement beneath a service report about Postgres (which was developed on SunOS and has been used in production functions on Solaris since about 2.5.1) might not be too swift. My guess, however, is that this was deliberate - a consequence of the services tail starting to routinely wag to Sun dog.

Media response, however, should not have been affected by internal confusion about agendas, so how many pundits or reporters got the real message?

Answer: apparently none, or, at least, none in the first 25 google news hits on "ZFS Solaris" as of Sunday November 27.

Some of the media write-ups verged on the comical: Todd R. Weiss writing for Computerworld said (among other things):


ZFS marks the first new file system for Solaris since 1981, when the UFF file system was developed. ZFS is available for OpenSolaris immediately and will be available for Solaris 10 next month through Sun's software maintenance programme.

1981 UFF for Solaris huh? Cool.

The best report by far came from Elizabeth Montalbano writing for Techworld through the IDG News Service. Here's the key bit from her report:


The new ZFS Solaris file system is not merely an incremental improvement over the OS's previous file system, but is actually a "fundamentally new approach to data management." ZFS presents a pooled storage model that eliminates the concept of volumes and its associated problems of partitions, provisioning, wasted bandwidth and stranded storage, according to the site.

ZFS allows thousands of file systems to draw from a common storage pool, each one taking up only as much space as it needs, while the combined I/O bandwidth of all the devices in the storage pool is available to all file systems at all times, according to the OpenSolaris Community. The file system also supports self-healing data through error detection and correction, and task automation that simplifies storage management, according to Sun.

So what are all these guys missing? They're missing the end of RAID storage as we know it and the spread of both Dtrace and ZFS to Linux and the BSDs.

Sun's Jeff Bonwick has a long blog on zraid - here are his last two paragraphs:


Which brings us to the coolest thing about RAID-Z: self-healing data. In addition to handling whole-disk failure, RAID-Z can also detect and correct silent data corruption. Whenever you read a RAID-Z block, ZFS compares it against its checksum. If the data disks didn't return the right answer, ZFS reads the parity and then does combinatorial reconstruction to figure out which disk returned bad data. It then repairs the damaged disk and returns good data to the application. ZFS also reports the incident through Solaris FMA so that the system administrator knows that one of the disks is silently failing.

Finally, note that RAID-Z doesn't require any special hardware. It doesn't need NVRAM for correctness, and it doesn't need write buffering for good performance. With RAID-Z, ZFS makes good on the original RAID promise: it provides fast, reliable storage using cheap, commodity disks.

Combine that with an almost ridiculous ease of administration and what you get is a real threat to companies like EMC, a triumph for open source, a significant, long term, Unix advantage over Windows, another key milestone achieved in the move to network computing, and a revolution in storage technology - but not a single reporter or pundit seems to have picked up on any of it.