back to article HDS brings out all-flash A series array

Hitachi Data Systems is taking its marketing attack direct to XtremIO and Pure Storage with a brand new, entry-level, all-flash array storage line, the HFS A series, separate from its existing VSP and HUS arrays. The HFS (Hitachi Flash Storage) product comes in three models: A220, A250 and A270, each with dual active:active …

  1. Anonymous Coward
    Anonymous Coward

    Thin provisioning is not a "saving"

    From https://www.hds.com/products/all-flash-storage/hitachi-flash-storage.html

    "5:1 average data reduction rates with data compression, data deduplication and thin provisioning enabled."

    Since when is thin provisioning a "reduction"? You're just presenting a server with more storage than it needs, not reducing what it actually writes to the array.

    1. Rob Commins

      Re: Thin provisioning is not a "saving"

      Agreed - Thin Provisioning should not be included in data reduction figures. Hell, why not TP 5PB and say you have a 98% data reduction rate? #overmarketing

      1. dikrek
        Boffin

        Re: Thin provisioning is not a "saving"

        Hi all, Dimitris from NetApp here.

        Indeed, thin provisioning is not "true" savings. Many vendors claim 2:1 savings from thin provisioning alone.

        Rob, agreed: It's easy to demonstrate 100:1 savings from that feature by massively overprovisioning.

        It is what it is, that's the state of capacity reporting these days and marketing claims. Most arrays are showing savings including thin provisioning, PLUS compression and dedupe where available.

        Check this out for some pointers on how to calculate savings and ignore what the GUI shows:

        http://recoverymonkey.org/2015/06/15/calculating-the-true-cost-of-space-efficient-flash-solutions/

        A lot depends on your perspective and what you're comparing the system to.

        I've seen a NetApp system for Oracle run at 10000% (yes ten thousand percent) efficiency since that customer was using an insane number of DB clones.

        If your existing system can't do fast, non-performance-impacting clones, then clearly comparing it to the NetApp system would mean NetApp would show as hugely efficient.

        If, on the other hand, your system can also do the fancy clones, AND thin provisioning, then in order to compare efficiencies you need to compare other things...

        Thx

        D

      2. Flammi

        Re: Thin provisioning is not a "saving"

        That's very funny. Tegile complaining about overmarketing. #companywithnoip

  2. Anonymous Coward
    Anonymous Coward

    With Pure, you can multiply IOPS x 32KB to get Bandwidth. Here, ... ? So in the IOPS tests, what size IO does HP use?

  3. Anonymous Coward
    Anonymous Coward

    Oops, in my last comment, I meant HDS, not HP, sorry

  4. Anonymous Coward
    Anonymous Coward

    HDS predicts AFA is a storage trend that will go away

    Can anyone say "backflip"?

    From HDS just over 12 months ago...

    Hu Yoshida's blog - Popular Storage Trends That Will Go Away: AFA and EFSS

    https://community.hds.com/community/innovation-center/hus-place/blog/2014/11/11/popular-storage-trends-that-will-go-away-afa-and-efss

    1. Anonymous Coward
      Anonymous Coward

      Re: HDS predicts AFA is a storage trend that will go away

      A lot of pretty basic features still roadmap items here and what happened to the all conquering HDS Flash Module Drives (FMD's) ? Looks like off the shelf MLC technology won that argument.

      1. This post has been deleted by its author

  5. Man Mountain

    I hate it when vendors talk about 'effective capacity' as if it's real - "With the A270 offering up to 384TB effective capacity in 2U, HDS claims it is the industry’s densest all-flash array." The only real thing is usable capacity, everything else your mileage may vary! And base your sums on 5:1 and you're really rolling the dice!

    1. Anonymous Coward
      Anonymous Coward

      All AFA vendors assume 5:1 dedupe so whether it's raw or effective, it's all relative and presumably still the most dense.

  6. Anonymous Coward
    Anonymous Coward

    Latency ??

    Flash is all about latency, in case you decide for it.

    And here they just claim "better than 1 ms". So this is not really good:

    IBM FlashSystem is in the range of 100 us.

    Wonder about a full market comparison here.

    1. Anonymous Coward
      Anonymous Coward

      Re: Latency ??

      IBM still sells storage? Clearly not very much from their last report.

  7. bmadaio

    HDSer Offers Thoughts...

    Biased HDSer here...

    Lot's of interesting (and some less interesting) points here.

    Thin provisioning? Sure, it's not reduction, per se, but other vendors are similarly including it in conversations, so I'm not sure we broke any real awkward ground there. Compared to old non-over-provisioned capacity, it's still a big win for many customers. However, if your concern is that your mileage will vary with this (agreed), we should talk about deduplication in the same dubious light. You know about averages, right? Either way, at 60 SSDs x 1.6TB each in 2U, we are comfortable with our capacity / rack unit, either effective or raw.

    Response time? Yes, we fully agree that response is as (probably way more, in most cases) critical to apps and customers than IOPS. This was a product announcement, not a technical whitepaper though, and anyone who puts real value on other vendors "minimum response time" numbers on the web should please email so I can describe some swamp land I have for sale. Call your local HDS rep/partner for more details, and of course, PoC things! (Oh, and for that IOPS number it was 4K block size.)

    On Hu's prediction, well I am not going to speak for Hu, but when you are selling customers flash by the petabyte and prices drop dramatically, you realize that things have changed. It is what it is. That said, most of us on this forum know that if only "need" was taken into consideration, Hu is right --> most environments could/should be tiered, especially if you have real time tiering like our HDT active flash mode. However, the benefit to all-flash is real and now much more achievable. I think even Gartner will concede (as Hu mentioned in his post) that eventually SSA/AFA and "general purpose storage" all serve the same need and will all be viewed in the same context. Maybe after a few more SSA stragglers die off or are bought for pennies on the dollar there will be fewer dots and Magic Quadrants can be merged. :-p

    RE: our FMDs, they are still going strong to the tune of 50PB+ last quarter and they still wipe the floor with your average SSDs. At 6.4TB each and built-in compression, the new FMD-DC2 removes challenges of overworked SSDs and takes compression off the controller. VSP F models gets you all flash versions with all HW/SW needed in a single SKU, if VSP is the solution for you. (Price one, you'll be happily surprised.)

    But if you want block dedupe today and a 2U solution (among other things), HFS is way to go.

    As we've all seen in the real world, there is more than one way to answer a problem and more than one type of problem to be solved. Much like houses aren't built built with only a hammer, nor are most IT infrastructure realistically supported with only one type of storage. This is largely the same reason grown up vendors used to have non-compatible midrange/high-end solutions. (oh wait, most still do...)

    Thanks for tolerating me!

    Bob

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2021