Home | GestaltIT | Enterprise Computing: New HDS AMS – Do We Need Enterprise Storage?

Enterprise Computing: New HDS AMS – Do We Need Enterprise Storage?

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 0 Flares ×

I’ve just been reading through the press release from HDS on their new AMS enhancements.  Hu Yoshida has blogged about the new features too (it makes a change to hear something different than another discussion on UVM).  There’s now HDP support as well as dense storage trays capable of holding 48 drives in 4U.  With all the new features, is there any need for enterprise arrays like the USP?

OK, I touched on this subject not that long ago when I criticised the recent USP-V clustering announcement.  Looking at what the AMS offers today, it’s not that far behind the enterprise models on the features it offers.  Let’s use Martin Glassborow’s (Storagebod) definition of Enterprise Storage:

  • Highly Available – 99.99%+ available
  • Supports multiple disk-types and sizes within the array
  • Supports multiple RAID Levels
  • Highly Scalable – Supports 500+ disks and supports many hosts attached
  • Highly Performant – Whatever that means
  • Non-disruptive upgrades – Internal code and hardware can be replaced/upgraded with no service outage

Does the AMS2000 range support all these requirements?  I think it pretty much does (although scalability for numbers of hosts/LUNs may be an issue).  Add in the new HDP features, active/active multi-pathing, dense storage trays and AMS2000 becomes a compelling purchase over standard Enterprise arrays.

So perhaps it isn’t necessary to purchase a huge USP-V for all your storage needs.  In fact, it may be prudent, depending on cost, to consider a USP-VM for tier 1 applications and the AMS range for tier 2 and below.  What do you think?

About Chris M Evans

Chris M Evans has worked in the technology industry since 1987, starting as a systems programmer on the IBM mainframe platform, while retaining an interest in storage. After working abroad, he co-founded an Internet-based music distribution company during the .com era, returning to consultancy in the new millennium. In 2009 Chris co-founded Langton Blue Ltd (www.langtonblue.com), a boutique consultancy firm focused on delivering business benefit through efficient technology deployments. Chris writes a popular blog at http://blog.architecting.it, attends many conferences and invitation-only events and can be found providing regular industry contributions through Twitter (@chrismevans) and other social media outlets.
  • Ron Major

    We are evaluating storage arrays to replace our current environment which consists of tier 1 and tier 2 frames. If it were not for our mainframe, tier 2 frames such as the AMS 2000 or XIV would meet our needs. Tier 2 storage has come along way in the last few years. I would say they are now where the tier 1 frames were a few years ago. In fact IBM is marketing the XIV as tier 1 at a tier 2 price.

  • Pingback: Cinetica Blog » HDS e i Modulari()

  • http://blogs.rupturedmonkey.com Nigel

    Chris,

    Interesting post.

    I think scalability has always been a major differentiator – enterprise scales well compared to midrange/modular.

    Of the announcements made re the AMS enhancements, Im not sure any of them change where the AMS stands relative to the items on storagebods definiton of enterprise.

    For me a major differentiator between the AMS and the USP has been around services like TrueCopy, ShadowImage and the likes. The AMS and its predecessors have always been severely limited in comparisaon to the USP (number of pairs and scalability of the software…….). I suppose the limited cache and ports is a huge factor and restricts number of pairs and the likes.

    Now if you could cluster AMS’s then that might be intereting.

    Now what does the 32MB page size tell us about TP extent sizes? How come Hitachi didnt go the route of other vendors and have an extent size measured in KB rather than MB? I Will be posting on that at some point.

  • Frank Bohnsack

    Hi Chris, we are already running most of our applications on AMS1000 systems; only about 20% of our storage is enterprise storage, most of it USP. Our tests prior to deploying the first AMS2000 system indicated that the performance has been considerably improved over the AMS1000. Add to this that more and more enterprise functions are added to midrange arrays and the application for enterprise storage becomes smaller and smaller. Major issues for me in midrange storage are:
    - Availability: It becomes more and more difficult (if not to say almost impossible in some cases) to take applications offline for maintenance work. So the ability to non-disruptive upgrade (icluding microcode) and repair is absolutely essential.
    - Attachment: The major impediment not to scale our current midrange systems beyond one frame is the limited number of ports these systems offer in association with the number of servers we need to connect and the number of LUNs we need to provision.
    I believe that midrange and enterprise storage will converge over the next years up to a point when we won’t deploy different kinds of arrays anymore but will have all tiers of storage in one system. There are already vendors offering systems with the same feature set in their midrange and enterprise arrays, the major difference between both classes just being the scalability (capacity and ports).

  • http://www.hds.com/ Mark Adams

    The line between enterprise and midrange storage has always been a bit fuzzy and evolves over time. What was true yesterday may not be true today and is even less likely to be true tomorrow. Customers that are making purchasing decisions are not just considering their immediate requirements but are looking out up to 5 years and need storage that will hold up well over time.

    Hitachi (where I work) is focused on bringing enterprise class capabilities (like dynamic provisioning) to the midrange products so they can meet the needs of the midrange market throughout their lifecycle. At the same time we’re also enhancing our enterprise products. The USP V/VM provide features the evolving enterprise market requires (storage virtualization, advanced data mobility, tiered storage management, massive scalability, high performance, etc.).

    You’re right in that we’re bringing the ‘enterprise-class’ functionality to our midrange AMS 2000. Many of these were previously considered “enterprise-class”:
    * Automatic allocation of workloads between controllers to alleviate bottlenecks
    * Eliminates administrative complexities of ensuring that servers have primary paths to the controller that “owns” their LUNs
    * Up to 32 backend data paths (controller to disk) for the best performance available from modular storage
    * Choice of high performance SAS and lower $$/GB SATA drives that can be intermixed in the same tray
    * Dense disk trays that reduce the amount of GB/floor tile
    * Dynamic provisioning that improves performance with wide striping and reduces up front purchases of extra capacity
    * Drive spin down to reduce energy costs

    As I like to see it, both the USP and AMS 2000 are optimized for the ever-evolving requirements of both the enterprise and midrange market.

  • Dave Brown

    We are UR’ing a copy of our mainframe Mainframe data from a USP onto an AMS2500 virtualized behind a USP-v. Tier 1 on USP-v. USP-v virtualizing AMS with FC and SATA disks. Giving us 3 tiers of storage on and behind a USP-v. Virtualizing EMC DMX3000, DMX4 and Clariions behind the USP-v. Gives us one skill set, one management interface and one replication process. It just don’t get any better than this.

  • Bill Ruehl

    As Nigel mentioned already, the key reason for the use of the USPV is the replication technology as well as the increased cache and ports. Tier II arrays just seem to be lacking in the asynchronous replication area. Currently any arrays that we are buying are going behind the USPV in order to use HUR if replication is necessary.

    The whole one skillset argument isn’t quite correct though,as you still need to present and layout the storage on the DMX,CX,etc in order to virtualize it behind the USPV, meaning you need the skills to allocate storage from those arrays as well as planning the layout on them.

  • Pingback: Wie viel Enterprise Storage steckt in der HDS AMS2000? | Dr.Storage()

  • Niels Roetert

    Chris,

    Isn’t one of the enterprise features, that during maintenance not half of your performance, capacity and availability disappears?

    -niels

  • http://www.brookend.com Chris Evans

    Niels

    Good question. Typically if a modular array lost a controller, you’d have that issue of write-through cache, loss of redundancy and so on. From memory, the AMS2000 doesn’t have that same issue as it is architected differently, so maintenance doesn’t have that problem.

    Chris

  • Chris Evans

    Ah Ron

    Good point – mainframe is effectively an “enterprise” feature….

    Chris

  • Chris Evans

    Dave, sounds good – no doubt that UVM is a great product. What are your tier 1/2/3 ratios?

    Chris

  • Chris Evans

    Mark

    Agreed. Perhaps the main differential is the scalability of each of the platforms, both performance and capacity.

    Chris

  • Chris Evans

    Frank

    Thanks for the feedback. I think you’re right that platforms are converging. We can see that already with EMC’s Symmetrix/CLARiiON and other devices. It makes sense really; the technology is becoming more reliable; customers want reduced cost and commodity storage will be delivered in the future using commodity components. The differentiator will be software.

    Chris

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 0 Flares ×