Background — High Performance Attached Processors Handicapped By Architecture
The application of high-performance accelerators, notably GPUs, GPGPUs (APUs in AMD terminology) to a variety of computing problems has blossomed over the last decade, resulting in ever more affordable compute power for both horizon and mundane problems, along with growing revenue streams for a growing industry ecosystem. Adding heat to an already active mix, Intel’s Xeon Phi accelerators, the most recent addition to the GPU ecosystem, have the potential to speed adoption even further due to hoped-for synergies generated by the immense universe of x86 code that could potentially run on the Xeon Phi cores.
However, despite any potential synergies, GPUs (I will use this term generically to refer to all forms of these attached accelerators as they currently exist in the market) suffer from a fundamental architectural problem — they are very distant, in terms of latency, from the main scalar system memory and are not part of the coherent memory domain. This in turn has major impacts on performance, cost, design of the GPUs, and the structure of the algorithms:
Performance — The latency for memory accesses generally dictated by PCIe latencies, which while much improved over previous generations, are a factor of 100 or more longer than latency from coherent cache or local scalar CPU memory. While clever design and programming, such as overlapping and buffering multiple transfers can hide the latency in a series of transfers, it is difficult to hide the latency for an initial block of data. Even AMD’s integrated APUs, in which the GPU elements are on a common die, do not share a common memory space, and explicit transfers are made in and out of the APU memory.
Dell just picked up Enstratius for an undisclosed amount today, making the cloud management vendor the latest well-known cloud controller to get snapped up by a big infrastructure or OS vendor. Dell will add Enstratius cloud management capabilities to its existing management suite for converged and cloudy infrastructure, which includes element manager and configuration automator Active System Manager (ASM, the re-named assets acquired with Gale Technologies in November), Quest Foglight performance monitoring, and (maybe) what’s still around from Scalent and DynamicOps.
This is a good move for Dell, but it doesn’t exactly clarify where all these management capabilities will fall out. The current ASM product seems to be a combo of code from the original Scalent acquisition upgraded with the GaleForce product; regardless of what’s in it, though, what it does is discover, configure and deploy physical and virtual converged infrastructure components. A private cloud automation platform, basically. Like all private cloud management stacks, it does rapid template-based provisioning and workflow orchestration. But it doesn’t provision apps or provision to public or open-source cloud stacks. That’s where Enstratius comes in.
I’ve noticed a growing trend among Asia Pacific organizations over the past 6-12 months: complete IT resistance to SaaS has steadily given way to more pragmatic discussions, even if IT has come to the table grudgingly. Over the next two years I expect this trend to accelerate. In fact, I believe that many SaaS solutions, particularly those that cross business and functional boundaries, will be rapidly subsumed within the broader IT portfolio, even if they were originally sourced outside IT.
Many SaaS vendors report already seeing more IT involvement in procurement, requirements definition, RFP creation, and negotiations. The clear procurement guidelines published by the IT department of the Australian Government Information Management Office (AGIMO) is one high profile example. Don’t get me wrong, in most instances business decision-makers will still lead, particularly in identifying the required business processes and determining how best to consume SaaS-based services. But IT decision-makers are getting more involved, particularly around integration.
Some areas to consider as you look to work more closely with business decision-makers to evaluate and negotiate SaaS and other public cloud deals:
No one would claim that the US tech market is booming. With Europe still mired in recession and debt problems, US economic growth looking soft, and business and consumer worries about the US government raising tax rates and cutting Federal spending, it is not surprising that businesses and governments are being cautious in their purchases of technology goods and services. But we think the fear is overblown. Forrester's forecast for the US tech market in 2013 and 2014 -- published today as "US Tech Market Outlook For 2013 And 2014: Better Times Ahead" -- projects a 6.2% rise in 2013 and a 6.8% growth in 2014 in US business and government purchases of computer equipment, communications equipment, software, IT consulting and systems integration services, and IT outsourcing. Adding in slow growing telecommunications services pulls growth down to 5.7% in 2013 and 6.1% in 2014. That may not be a boom, but it is certainly not a bust.
While CIOs are cautious in their tech buying -- and in the case of the Federal government, actually cutting back -- that caution has and will show up mostly in reduced spending on computer and communications equipment (with the exception of tablets). CIOs will be most aggressive in software, especially for SaaS apps, analytics, and mobile apps. IT outsourcing will see good growth in 2013 as the result of 2012 selection decisions, while IT consulting and systems integration will come on strong in 2014. Business and government purchases of telecommunications services will continue to grow at a slower rate than the overall tech market.
As businesses get serious about the cloud, developers are bringing more business-critical transaction data to cloud-resident web and mobile apps. Indeed, web and mobile apps that drive systems of engagement (how you interact with your customers and partners) are the reason why many companies look to the cloud in the first place. Public clouds offer the speed and agility developers want, plus the development tools they need. Once you’ve built a killer web or mobile app in the cloud and it’s in production, driving real revenue, who’s responsible for making sure it performs?
It’s a team effort. Developers have to think about performance management as they build, and IT operations teams need to design application monitoring and management into their cloud deployment processes up front. Why? Because there’s no time to do it later. You won’t have time to implement a new app monitoring solution for each new cloud app before you need to get it out to users. And once it’s out there, you need to be tracking user experience immediately.
In traditional IT, one of the reasons we could get away with limited insight into application performance was because we usually overprovisioned resources to make sure we didn’t have to worry about it. It’s easier to have excess capacity than to solve tricky performance problems – problems you might only see once in a while.
Every company generates data that would be of significant value to its customers, partners and potential partners; information that could be combined with insights from this ecosystem, public data and other sources to generate significant new discoveries, products and business values. But making our data available, easily consumable and getting payback for sharing it are significant hurdles.
Over many years we have built up an ever-more complex web of security, legal and data management practices that make it nearly impossible to share valuable info between companies in an open marketplace style – which is exactly what is needed to open up this value.
But it doesn’t have to be this way. There is a new approach that leading enterprises and governments are taking today that is significantly simpler, more manageable and empowers companies to share their key data more freely, opening up massive new market opportunities for all. Here's how a few Forrester clients are taking advantage of this new model:
At the OpenStack Summit in Portland last week, the open-source cloud platform got real, to echo Forrester’s cloud team predictions for 2013. At the busy gathering attended by over 2,400, suits mingled effortlessly with hoodies and deep-tech design committee meetings were sandwiched between marquee-name customers sharing success stories. Three core themes drove the show, as outlined by Jonathan Bryce in the opening keynote: the OpenStack technology platform has matured, the ecosystem is vibrant, and the global user footprint now includes enterprise customers doing real business.
The show followed on the heels of the Grizzly release, the 7th release of the OpenStack platform. Along with stronger support for VMware and Microsoft hypervisors, Grizzly widens block storage options and includes 10+ new enterprise storage platform drivers and workload-based scheduling. A wide range of new network plugins expand the platform’s software-defined networking options and a sexier Dashboard to access, provision and automate resources.
The OpenStack Foundation and Microsoft have released major updates to their cloud platforms and frankly there’s nothing really new or exciting here – which is a good thing.
Sure, there were over 250 new features added in the Grizzly release of OpenStack that brought several nice enhancements to its software-defined networking, storage services, computing scalability and reliability and it delivered better support for multiple hypervisors and better image sharing, too. The vSphere driver was given a significant update, Swift got better monitoring, and there's a new bare metal provisioning option, which was the talk of day one of the OpenStack Summit here in Portland, Oregon.
For Microsoft, it lifted the preview tag from its full Infrastructure as a Service (IaaS) enhancement to the Windows Azure public cloud platform. It’s a big deal for Microsoft who previously didn’t provide this level of virtual infrastructure control but compared to the rest of the public IaaS market, it’s more of a “welcome to the party” announcement than a new innovation or differentiator. To sweeten its appeal, Microsoft added a pledge to match AWS pricing for compute, network and storage services and thus dropped its prices in these areas by 21-33%.
HP today announced the Moonshot 1500 server, their first official volume product in the Project Moonshot server product family (the initial Redstone, a Calxeda ARM-based server, was only available in limited quantities as a development system), and it represents both a significant product today and a major stake in the ground for future products, both from HP and eventually from competitors. It’s initial attractions – an extreme density low power x86 server platform for a variety of low-to-midrange CPU workloads – hides the fact that it is probably a blueprint for both a family of future products from HP as well as similar products from other vendors.
Geek Stuff – What was Announced
The Moonshot 1500 is a 4.3U enclosure that can contain up to 45 plug-in server cartridges, each one a complete server node with a dual-core Intel Atom 1200 CPU, up to 8 GB of memory and a single disk or SSD device, up to 1 TB, and the servers share common power supplies and cooling. But beyond the density, the real attraction of the MS1500 is its scalable fabric and CPU-agnostic architecture. Embedded in the chassis are multiple fabrics for storage, management and network giving the MS1500 (my acronym, not an official HP label) some of the advantages of a blade server without the advanced management capabilities. At initial shipment, only the network and management fabric will be enabled by the system firmware, with each chassis having up two Gb Ethernet switches (technically they can be configured with one, but nobody will do so), allowing the 45 servers to share uplinks to the enterprise network.
Digital capability – social, mobile, cloud, data & analytics – disrupts business models, introduces new competitive threats, and places new demands on your business. Highlighting this fact: Forrester’s 2012 “Digital Readiness Assessment” survey found that 65% of global executives say they are “excited about the changes that digital tools and experiences will bring” to their company.
While most people know these digital trends are coming, however, far fewer know how to purchase these cutting-edge digital capabilities. What companies will you rely on? Where are the new risks? What are the pricing models? In the survey mentioned above, only 32% of the same sample agreed that their organization “has policies and business practices in place to adapt” to those digital changes.
This is important, since developing the breadth of digital capabilities your company needs cannot all be done in-house. To succeed, your company will need to access the strengths of its supplier ecosystem, maximize value from strategic partners, and leverage emerging supplier models.
This is a tremendous opportunity for sourcing and vendor management professionals to increase the strategic value they provide to their business. But to do this, you’ll need to balance your traditional cost-cutting goals with demands for business expectations for growth, innovation, and value.