Lets face it, we'd all love a maid!

By | September 25, 2007

BOOM!  Hitachi has dropped another slipper by introducing their new MAID, and from a distance she looks quite promising 😉

Ok so I promise no more innuendos.

Basically, Hitachi have introduced to their AMS and WMS storage arrays the ability to designate RAID Groups as either “power saving” or “normal spinning”.  These new “power saving” RAID Groups can be spun down and spun up on demand although exact details of how this is accomplished are a little thin at the moment.

Now this is a technology that I think, if implemented well, is tailor made for at least two situations –

  • Backup
  • Archive

I know I know!  All the press releases have already listed these as ideal candidates but let me tell you why I will also be carrying the torch on this one…….

You will no doubt recall the recent blogsphere chatter concerning the astronomical numbers being quoted for UVM on the USP-V and the argument that the USP-V front end may not actually be able to support I/O to that much storage…..  Well I decided to take a look at some stats for an AMS1000 that was externalised behind a USP and saw some interesting numbers.  The vast majority of space on this AMS was used for email archiving and staging backups.  What I saw was that almost all of the RAID Groups were idle (zero I/O) for massive amounts of the day.  Im talking hours and hours of absolutely no activity.

Now this almost saw me pen a response to the FUD flingers but I never got around to it.However, at the time I also thought that these disks could do with being spun down when not in use.  And what do you know, a week or so later and Hitachi announce that they will support this.  Magic!

Now before I get carried away I would like to know more about how it actually works and certainly a little on how this will affect disk reliability.  

There are two things that stick in my mind re spinning disks up and down –

  1. Bill Watkins, CEO of Seagate, saying that he never powered down his home storage in response to being asked about the reliability of disk electronics when constantly powered on and off.  He should know
  2. Working for a company that went through some major UPS and power work at its primary date centre.  We had to power the data centre down every Sunday evening for several weeks.  Each week on powering our EVA storage arrays back up we, almost without fail, had at least one failed drive (FC).

With these in mind, despite all the jazzy stuff on modern hard drives (parking the heads off of the platter and improved electronics on the circuit board…….) I’m still sceptical about spinning drives down and this is not a fear I am going to overcome over night.

Like I said at the start, this looks interesting from a distance, but so do a lot of people things.  We’ll have to wait and see how it measures up on closer inspection 😉

Nigel

12 thoughts on “Lets face it, we'd all love a maid!

  1. snig

    This is a pretty big differentiator from the other modular systems out there.  I wonder how the competition will try and spin it….

  2. Storagezilla

    Well to be fair he does say RAID-6 with SATA, one thinks that recommendation probably does extend to FC as well to be sure to be sure.

  3. TimC

    Glad you brought up the reliability thing Nigel, because that was my first thought as well.  The one thing Hitachi DOES have going for it is that they control those hard drives too… here's hoping they have some technology in their back pocket for their sata drives that will allow graceful power-up and power down.  To me, that seems like the biggest issue.  If you can let the drive know a power down is coming, and get it into a graceful state, one would think it shouldn't be *SO BAD*.

    And this could be a boon for the entire industry if they pull it off.

  4. Nigel

    Snig I agree that if done well this could be a huge differentiator…. until the competition catches up of course!

    Tim, I feel that the drive reliability issue is absolutely key to the success of this.  If done well this could be huge for Hitachi and its competitors…. and who knows it may even do bit for the environment too.For a change I see this as being a green initiative with a bit of meat on the bones.

    Bring it on.

  5. the storage anarchist

    First, I gotta wonder if Hitachi is even using Hitachi SATA drives in the AMS. I know first-hand that they ship an awful lot of non-Hitachi Fibre Channel drives in their USP and USP-V systems. Not sure why they don’t use their own drives, but I’ve seen ’em and touched ’em, and they don’t all say "Hitachi."
     
    On the other hand, I wonder if they’re still doing read-after-write on their SATA drives, as Hu previously boasted back when the AMS support for SATA was first launched. If so, the added workload on those drives are not only reducing performance, but ADDING to their power profile – potentially negating any of the porported  MAID-related savings.
     
    Or maybe the quoted 20% power savings is because they finally stopped doing the read-after-writes all together. Less power, but higher risk of data corruption – interesting trade-off. But it might just put the AMS power back in line with competitive SATA products – you know, the ones with signiifcantly larger market share, like CLARiiON.
     
    Smells like Hitachi Math to me! (see http://itknowledgeexchange.techtarget.com/overheard/overheard-hitachi-math/)
     
    Well…you ASKED me to respond, now didn’t you?

  6. Newbie

    >maybe the quoted 20% power savings is because they finally stopped doing the read-after-writes all together

    I'm not sure that "removing read-after-writes" can transform into "power down disks" even in the world of marketing. So, guess it's an actual power down, not a trade-off…>[Power Saving] can be invoked by clients or applications as a storage service on an as-needed basis

    To me this one looks as ability not only power-down via array-config utility but also via server-side script or agent… Sounds neat…

  7. Newbie

    That’s basically what I did but posts still get messed…Another try?… Or may be it’s just firefox….

  8. Nigel Poulton

    Hmmmmmm don’t remember asking you to comment Barry???

     

    Re the non Hitachi drives in the USP, that’s no secret, you will see plenty of Seagate drives in HDS storage.  Some offer features that other don’t have such as fast format etc…  Never seen an EMC drive in a storage array though.  Funny that, considering EMC is supposedly where information lives.  Ive personally seen information living on Seagate, Hitachi/IBM, Toshiba…….. but never EMC.  Strange?

    May be they will add technology to their drives and recommend only using Hitachi drives as power saving drives until others catch up?  Who knows but the possibilities are quite exciting.

    And dont for one minute say you wont be offering a variation of the same yourselves if this proves useful and catches on.

    Oh and I think I remember asking for your input over on your blog. 😉

  9. Newbie

    I recall a customer saying EMC should finish their slogan (after major array failure): "Where information lives… and dies". No offense, just funny stuff.

Leave a Reply

Your email address will not be published. Required fields are marked *


*

You can add images to your comment by clicking here.