HDS/Hitachi I believe were an investor in BlueArc, now they’ve acquired them.  I think this is a good move for both companies for a number of reasons.  Firstly, it marks the continuing change in HDS’ strategy, which now includes acquiring companies, albeit companies that they have an existing relationship with.  This approach may seem cautious but perhaps it’s a good thing to take time to understand the people, technology and culture, although HDS could have made their move earlier.  Second, I believe BlueArc suffers in a market where (large enterprise) customers expect longevity.  Committing to a new platform that suddenly has little or no development could be a costly mistake.  There’s also the issue of global support and maintenance to consider too.  Being part of Hitachi will give customers more reassurance.

Some time ago, I put the following diagram together as part of work for a customer.  it was meant to show how the Hitachi technology can operate on a number of layers, from the physical hardware upwards.  Although I have no specific inside knowledge, I expect Hitachi will converge the physical layer into a single device at some stage.  Chris Mellor’s article hinted at this (although I expect this slide is no longer accurate) and developments in their technology imply this too.  The AMS2XXX range certainly became more USP-like in its last release.

Operating a single hardware platform makes sense as HDS can abstract the hardware from the logical view using tools like HTSM and UVM.  Presentation is then through native FC/FCoE or using gateway products to provide NAS and object connectivity.  VSP represents a much more scalable architecture than the USP was, one of the key features being the abstraction of the front and back-end processors from the physical interfaces.  This scalability easily supports multi-protocol shared environments.

Other vendors have gone down the route of doing scale-out through multiple interconnected nodes.  I think this is a good approach and it could be argued that Hitachi have taken another way and gone for single monolithic arrays.  I don’t believe this is the case as the VSP is already a cluster of two nodes and from the PCIe backplane architecture can presumably scale to more nodes as required.  They are however more closely coupled than other implementations.  If Hitachi delivers on the potential promise of HAM, then clustering/nodes can be implemented and the single monolithic array isn’t an issue.  Now that Hitachi own BlueArc, they can integrate the functionality to ensure any NAS presentation can be mapped back to any storage array, with seamless migration.

So in summary, the future looks good, I’m looking forward to seeing how Hitachi capitalise on their new investment.

Disclaimer: I’ve been engaged with HDS as a blogger, including trips to Japan and the US.  I have worked for HDS UK delivering consultancy to their clients.

Since being discontinued, HP has decided to hold a fire sale, selling off the 16GB Touchpad at $99.  I’ve tried to get hold of a cheap model in the UK, so far without success as the reduced price point has been incredibly popular.  Should we be surprised? Of course not.

Beating Apple

Any tablet competition coming into the market needs to pitching itself against Apple and the iPad.  This is no mean feat to achieve, as Apple hold the dominant position in consumer electronics today, with a fanatical fanbase that will lap up almost any new device they produce.  I was originally skeptical about the iPad, seeing it as a limited usage device, which for me it has turned out to be.  My original iPad v1 is really an expensive ebook reader.  Yes, I use it for other things, but there’s pretty much nothing I can’t use my iPhone for, unless of course I need the benefit of the larger screen.  For “power” work, I use my MacBook.  However, irrespective of my own personal hangups, it’s clear that the iPad and tablets in general have  developed into a new market segment, with most consumers happy to pay the Apple iPad premium cost.

HP of course were positioning themselves at the Enterprise, selling the virtues of a platform that could securely display Enterprise data and act seamlessly from PC to tablet to smartphone.  Within the Enterprise, customers can be price sensitive or agnostic, depending on the purpose.  This may seem like I’m contradicting myself, but imagine senior executives in a large organisation; they will demand access via iPad and price will be irrelevant.  For large scale deployment and adoption across the Enterprise, however, price is a key sensitivity.  I’d suggest HP was targeting the second group, so pricing their new-to-market, device at a similar position to the iPad  was a clear mistake.

Getting a Bargain

What we’ve seen with the HP “fire-sale” of the TouchPad is that everyone loves a bargain. For consumers unwilling or unable to pay Apple iPad prices, the draw of a $99 tablet is irresistible and so we’ve seen huge demand at this price-point, even though there are few (if any apps) and no statements on future support.  Does this mean HP could have adopted a better strategy?  The initial view might be to have sold the TouchPad at a loss in order to gain market share.  At $99, this would clearly have been a success (as we can now see) and this follows the model Microsoft followed with the XBox, selling the console at a loss against the PS2/3 in order to gain market share.  But there are a number of differences in the market between games consoles and tablets.  Firstly, as far as I am aware, all console manufacturers lose money or at least break even on the hardware.  This is because they own the development platform and charge a premium for game development, taking a cut on all games sold.  The losses of the hardware are made up on profit from the software.  In the tablet market, this isn’t the case.  Apple have to make profit on the hardware as so many apps are sold for free or very low cost, unlike high priced console games.  If HP chose to sell the TouchPad at a loss, they would have no opportunity to make that money back in software as they would have to make app development on the platform free to attract developers.

It appears that HP were in a lose-lose situation with the TouchPad.  In order to gain market share, they would have to accept selling the device at a loss for some time to come, until adoption levels were high enough to either increase the price or charge more for software.  At the same time they would have been up against the Cult of Apple, so it’s understandable that they chose to kill the tablet off so early.  What I don’t understand is why they went down the tablet route in the first place.

The Future

I wonder now what will the future of TouchPad and webOS will be.  There’s clearly a market for cheap tablet computing, the question is whether the devices can be produced at a price point to move people away from Apple.  The Android market beckons; they’ve been successful in acquiring customers who don’t like the Apple Content Control machine.  So perhaps HP should package and sell off the Touchpad to Google.  After all, they’ve just moved into the mobile hardware business and might jump at the chance of being able to beat Apple at their own game.

At previous HP events last year I noted that the recent storage acquisitions (Lefthand & Ibrix) were being powered by HP server hardware.  This makes sense as HP have a strong (read #1 market position) server product range.  This week HP revealed Store 360, the strategy to place all of those technologies on a consistent operating system platform.

What benefit does this give?  Well there are positives for HP and for customers.  From HP’s perspective they are simplifying their product development; lowering costs in the process.  Having the technology on a single platform enables software features to be migrated between the products; new features can be developed once and integrated across the entire product range.

For customers the benefits may be less obvious but they are there.  Standard HP hardware (including blades) can be used to deploy storage functionality and there’s the ability to re-purpose that hardware with no additional work.  So imagine in a private cloud environment the need to create additional file services or perhaps de-duplication.  An array can be build from an HP server & disk or even a single server blade.  The blade solution is already available today – the P4800 (and I have a video on that today).

It’s worth looking at the specific products to understand how arrays deployed on a common hardware platform could scale.  Lefthand, Ibrix and 3PAR are all node/clustered architectures that grow through adding additional nodes.  This fits perfectly with the concept of commodity or standardised hardware, something I discussed recently when talking about Nexenta.  The 3PAR platform may present more difficulty to commoditise as the nodes still use custom ASICs for on-the-fly thin provisioning functions.  Presumably these features could be provided by hardware on PCI-E cards or software.

At this stage Store 360 is a strategy, with products yet to be delivered.  However it’s the foundation to providing customer flexibility and for HP, increased competitiveness.  In line with previous comments I’ve made on the future of storage arrays, HP’s strategy confirms the hardware is less relevant and the features in software are most important and that’s where HP’s focus will be.

After discussions in the Bloggers’ Lounge and a backstage tour of the main keynote stage, 4pm saw the keynote speech of the day from Leo Apotheker, HP CEO.  This is the guy that looks like Patrick Stewart, talks like Arnold Schwarzenegger and came across as solid and knowledgeable about his products and strategy.  HP’s TouchPad was prominent throughout the session with the emphasis on how it is more suitable for the Enterprise than (presumably) the iPad.  Of course with consumer devices the details matter and the level of usability can’t be gleaned from a canned presentation.  I hope to find the TouchPad stand later today for a review.

The keynote concluded with Don Tapscott, whose insights into technology were thought provoking but possibly came too late in the day to be absorbed fully.  In summary the keynote was good but perhaps lasted too long and the quality of the presenters dipped in places.

By 7pm with keynotes over the Discover Zone was buzzing with 10,000 attendees taking advantage of the “free” food and drink.

Summary

Headline presentations are designed to rouse the troops and get everyone fired up; HP certainly achieved that.  As day 2 begins, it’s time to take stock and look at the substance behind these new announcements.

You can find pictures of HP Discover 2011 on Flickr or via The Storage Architect Facebook page.

• http://www.facebook.com/TheStorageArchitect
• http://www.flickr.com/photos/33818355@N07/sets/72157626902117768

Disclaimer:  As a blogger/attendee, HP paid for my flights, hotel and most meals.  I have not been compensated for my time and am not obliged to blog on the event.  All the opinions are my personal views.

The difficulty for the week will be determining what sessions to attend – there are over 800 during the next 4-5 days ranging from subjects including Cloud, storage, servers, networking, systems management and applications.  There’s also the partner pavilion, a showcase of products partner sponsors.

So all in all, lots to see and do, as well of course, as being in Las Vegas.  So for fun, here’s a video to start, showing the journey from my room to registration at the Discover event, which takes me about 10-15 minutes!

Click here to view the embedded video.

Blog Roll

• Rich Dunbar – http://webosroundup.com
• Chris M Evans – http://www.thestoragearchitect.com
• Mauricio Freitas – http://geekzone.co.nz
• Josh Greenbaum – http://ematters.wordpress.com
• Thomas Jones – http://www.niketown588.com
• Jason Kenney – http://itmanagecast.blogspot.com
• Charles King – http://www.pund-it.com
• Michael Krigsman – http://www.zdnet.com/blog/projectfailures
• Jake Ludington – http://www.jakeludington.com
• Stu Miniman – http://blogstu.wordpress.com
• Matt Norwood – http://www.insearchoftech.com
• John Obeto – http://absolutelywindows.com
• Devang Panchigar – http://storagenerve.com
• Nigel Poulton – http://nigelpoulton.com
• Hector Russo – http://geeksroom.com
• Jeffrey Powers – http://www.geekazine.com
• Philip Sellers - http://tech.philipsellers.com
• Eric Siebert – http://www.vsphere-land.com
• Enrico Signoretti – http://juku.it/en & http://juku.it
• George Salcedo – http://webosroundup.com
• Hans Vredevoort – http://www.hyper-v.nu
• Chris White – http://www.neowin.net
• Alex Williams – http://www.readwriteweb.com/enterprise
• Brandon Wirtz – http://blackwaterops.com

EMC World 2009 - Flash Price Erosion

In this image I took from Joe Tucci’s keynote speech at EMC World 2009, you can see the expected price erosion.  Despite this, flash still hasn’t been widely used.

So what have EMC said this week?  The flash announcement is more of a statement of direction than details on a specific product.  EMC will bring MLC flash to their products alongside existing SLC drives.  The main differences between these two technologies being cost, performance and reliability.  MLC flash tends to get used in consumer devices and so it’s really not up to a 24/7 access profile.  EMC are not the first company to do this however; Violin Memory have an MLC based product, the 3140 array.

EMC did make an announcement on a future product, currently codenamed Project Lightning, a PCIe server-based flash card.  Many of you will have read (and be aware of) Fusion-IO from this post (http://www.thestoragearchitect.com/2010/10/28/storage-networking-world-europe-ii/).  Again, EMC’s offering is not a first for the industry as Fusion-IO have been selling their flash cards (in a variety of formats) for some time.

What’s potentially different about Project Lighting is the integration into the server and virtualisation layer.  I’ve only Chad’s rather speculative post to base my thoughts on, but the idea appears to be to use server-based flash in a tiered storage model to improve VMware performance.  It helps to own the company of course, as you can direct the host developments towards matching the hardware developments; I can’t imagine many other array vendors have tried to run vSphere in the hardware, for example.

But where does Project Lightning take us to?  Well, it reinforces the view that there’s a direct dependency between server virtualisation and storage.  It means we’ll see hybrid storage/server solutions that make the best use of resources without having to deploy dedicated (and sometimes slightly siloed) technology towers.  There’s also some thinking required around the way this kind of infrastructure is deployed.  For example, in a traditional model of data replication, synchronous replication allowed the host data to be duplicated to another site.  If the primary site is lost, the remote copy of data can be accessed and appears to be a “crash” copy, which should recover successfully.  Virtualisation changed that and made it more difficult to replicate at the array level.  What now happens when we’ve a dynamic infrastructure that replicates data between storage/cache in hardware and disk in the server?  Where’s my data if any part of that infrastructure crashes?  These issues aren’t unsolvable, but do require more thought in terms of architecture design.

Assuming EMC can bring Project Lightning to market quickly, I expect they will see competition from HP, who have already approached the server/storage combination with their P4800 storage blades, their partnership with Violin and their work on memristors.  Whether we like it or not, the big vendors are moving us closer towards a converged world.

First of all, from an availability and resiliency standpoint, it’s worth having a look at Amazon’s definition of regions and availability zones.  AWS is currently available in 5 regions classed as; US East (Northern Virginia), US West (Northern California), EU (Ireland), Asia Pacific (Singapore), and Asia Pacific (Tokyo).  Within these regions there are multiple “availability zones” – separate locations which Amazon claim are “engineered” to be insulated from failures in other availability zones, presumably as physically separate data centres with independent networks, power delivery and so on.  It seems on the face of it reasonable to assume that if Amazon claim resiliency within a region by using availability zones that designing an infrastructure that sits in a single zone should be acceptable; I disagree.

As far as I am aware, Amazon publishes no specific details on how their infrastructure is plugged together and inter-operates across geographic boundaries.  Therefore it’s impossible to understand how availability zones actually work and how they have been engineered to isolate against failure.  As we saw yesterday, the whole of region US East was affected (and at the time of writing still is) regardless of location, making it obvious that the availability zone protection isn’t guaranteed in all circumstances.

When organisations design their own data centres, they understand their business requirements and the infrastructure is based on that information, including how and where data centres should be sited.  Financial organisations, for instance, are required to site their data centres a certain distance apart for resiliency.  Features such as synchronous replication at the array level, high availability, application data replication can all be used to ensure service is not disrupted because the infrastructure team have (hopefully) engaged with the application owners to understand their specific requirements.  If that requirement were (for example) 100% data integrity, then data would need to be synchronously replicated to another location to ensure it could be accessed in a recovery scenario.

Amazon have, with AWS, provided generic infrastructure without publishing specifics on how that infrastructure is delivered.  This is fine, as AWS is delivered as a service, however availability zones are not guaranteed against all failures (merely engineered against it) and it would be foolish to assume any organisation could guarantee against all possible disaster scenarios.

If you are delivering a service using cloud infrastructure it is your responsibility to determine the level of failure you are prepared to accept.  That could mean running services across multiple providers, a subject I discussed 2.5 years ago in this post; http://www.thestoragearchitect.com/2008/12/16/redundant-array-of-inexpensive-clouds-pt-ii/.  Although this post was more storage focused, the concepts still apply to application design.  If you’re starting a business from scratch, then there’s no excuse these days not to engineer across multiple regions or even multiple providers (in fact, the effort of going multi-region will be comparable to that of going multi-provider).  Obviously some applications will be more difficult to implement in a diverse manner than others, however looking at the four web-based applications I quoted at the top of this article, I expect that all of them have a large degree of read-only traffic and a lot of “write-new” data with only a small percentage being updates.  That being the case it would be relatively easy to distribute read I/O geographically and to stage writes in the same manner, synchronising data on a periodic basis.

The Storage Architect Principles

• Basing your infrastructure in “the cloud” is not a bad thing to do
• You must understand your business service requirements and design to them
• You must understand the service offering of the cloud provider
• Design around availability, resiliency,  and therefore mobility, at the application layer
• Using multiple providers is a good thing
• Don’t let cost saving blind you to reducing service quality

There’s one other thing to bear in mind (as the final bullet point above alludes to); US East is also the cheapest location for Amazon services (and I presume the largest).  The cynic in me wonders is some of the service implementations have been based on cost rather than service level availability, especially where these services are free to the end user.