Seagate's visible firmware problem

Seagate's visible firmware problem

Summary: A recent problem with the firmware in a Seagate drive offers a peek at the 400,000 lines of spaghetti code inside a modern disk. It is a cautionary tale for those who blithely assume that disk drives "just work.

SHARE:
9

A recent problem with the firmware in a Seagate drive offers a peek at the 400,000 lines of spaghetti code inside a modern disk. It is a cautionary tale for those who blithely assume that disk drives "just work."

In an article titled Seagate's AAK firmaware thoroughly examined an anonymous writer profiles reported performance problems with Seagate's 7200.10 drives running AAK firmware.

Long story short, the author found that the AAK firmware reliably ran 20 MB/sec slower than the same drives with different firmware, along with some other anomalies. Seagate has so far refused comment, but reportedly has replaced some of the drives.

Inside a modern disk drive Back in the mainframe dark ages, host CPUs directly controlled read/write heads. As chips got cheaper and faster more and more intelligence migrated to the disk.

Today a disk drive CPU is faster than a 1980's superminicomputer. Some of the disk's code dates back to the 1980's as well.

Architecture by accretion Disk drive bugs are no secret inside the industry. Array vendors test every version of disk drive firmware before they allow it inside their arrays. Why? Because all disk drive firmware has problems that can hose data. Or, as in this case, performance.

As of 6 years ago - and I doubt it has changed since - no disk drive vendor had done a complete re-write of their 400,000 lines of code. So new functionality grows by accretion, not architecture.

The Storage Bits take Seagate makes good drives. I'm taking delivery on one today. Yet any product with 400,000 lines of code will have bugs, and disks are no exception.

In this case the bug affects performance, something that can be measured. Too often the bugs affect data integrity and error handling (see How Microsoft puts your data at risk) which are harder to see but even more important.

We need file systems that ensure end-to-end data integrity. Only the file system knows what data is written and can compare it to the data that disk drives return. Other than ZFS though, it appears those file systems are years away.

If you rely on your computer for work or play, do yourself a favor and back it up at least monthly. The bugs are out there. Are you ready?

Comments welcome, of course.

Topics: Software, Hardware, Storage

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

9 comments
Log in or register to join the discussion
  • ZFS

    I don't know if you have watched this video. Jeff Bonwick and Bill More are presenting ZFS at BayLISA. (I saw them live at a different event. It was really great!!!)

    http://video.google.com/videoplay?docid=8164805665567899891&q=ZFS

    At around 1:17 they are talking about Disk Firmware. (Slide 13 http://www.opensolaris.org/os/community/os_user_groups/losug/zfs-technical-jeff-bonwick.pdf )

    Really interesting talk...
    Burana
  • Comments from article author

    Hi, I've written the article. I have a few comments:

    - Not only sequential performance was affected, since extracting 108,000 very small files to the same disk was quite slower on AAK disks. This is highly non-sequential I/O.
    - Sure, firmware is complicated, but STR (Sequential Transfer Rates) is easy to measure. Before releasing firmware into the wild, i hope Seagate, like any manufacturer, tests it. After all it's about the data of their customers and a breach of trust can have profound implications to their reputation and, ultimately, sales. It's unacceptable for disks with the same model number to feature much lower STR.
    - The AAK firmware probably was never meant to be sold in the retail market but rather made for a custom client, optimized for server-related I/O. Some of the benchmarks agree to this - the AAK firmware sure has its strengths. But i do not think it's suited for home users.
    - Sure dataloss bugs would have been much more severe, but for Seagate a performance issue can be just as devastating. After all, people who read about the firmware problems will simply choose a competing brand.
    - Aside from the technical issues, Seagate leaves it's customers in the dark by not revealing detailed information. I can only suspect that this is deliberate.

    All in all, I'm very disappointed in Seagate. I think they handled this poorly, especially from a marketing/sales point of view.

    P.S. there's a typo in the link, firmaware should be firmware.
    Fluffles.net
  • How can you tell...

    if your hard drive has this firmware?
    High Sierra
    • Read the article ;-)

      Read the article, page 6:
      http://www.fluffles.net/articles/seagate-AAK-firmware/6
      Fluffles.net
  • Shouldn't they use formal methods...

    Shouldn't they use formal methods (formal verification and validation) to mathematically test their code? Like NASA and others critical-systems designers are supposed to use.

    I mean, digital data is critical in this new digital world, they should provide bug-free firmware, isn't it?

    What would happen to our civilization if paper books just start to corrupt their content at random or spontaneously burn? I'm sick of the current reliability level of optical and magnetic media.

    Regards,

    MV
    MV_z
    • I agree

      They should. And maybe someone who wants to take over a big chunk of the disk
      market will invest the years to do so.

      As for the risks in electronic media see my post <a href="http://
      storagemojo.com/2006/06/19/reprint-jonathan-livingston-seagull-every-5-
      years/" >Reprint "Jonathan Livingstone Seagull" every 5 years?</a> for some links
      to digital preservationists.

      Thanks for writing.

      Robin
      R Harris
    • RE: What would happen to our civilization if paper books just start to ...

      corrupt their content at random or spontaneously burn?

      Well books do deteriorate over time, unless special efforts are used to preserve them.
      mrlinux
  • RE: Seagate's visible firmware problem

    "We need file systems that ensure end-to-end data integrity. Only the file system knows what data is written and can compare it to the data that disk drives return. Other than ZFS though, it appears those file systems are years away."

    Not really - in fact, at least one such file system has been around (and in wide-spread use) for quite a while: WAFL. It uses in-parent metadata in a manner conceptually similar to ZFS's to ensure that when a child is read it contains the actual data that it should, and is thus 'end-to-end' from the viewpoint of the kind of disk errors that you describe.
    - bill
  • RE: Seagate's visible firmware problem

    I have contacted the BBB Better Buisnnes
    and they have taken my case I have a Seagate sata 500gog 7200.11 that has gone to sleep as it where its OEM OMG if you know what I mean : Seagate dont give you any warranty at all !!! leaving me with a $900 dollar PC that just sits there, wont boot... Hp dont want to know because my 1 Year HP warranty has expired they say get a new $80 dollar warranty is the best that they offer. Does anyone know if there is a Class Action suit is on the Horizon, this firmware Problem is world wide could affect 10's of thousands of drives, and Seagate have not released a Recall ::
    7200.11 drives are still selling on shelfs ,, worse still they have withdrawn the Firmware fix,, it dosent work either. Can you imagine a worse situation for a Hard Drive company than this????
    andeemac2006@cox.net
    andeemac2006