How many smartphones are there in your household? How about laptops, tablets, PCs? What about other gadgets like Internet-enabled TVs or smart room temperature sensors? Once you start to think about it, it’s clear that even the least tech-savvy of us has at least one of these connected devices. Each device is constantly sending or receiving data over the Internet, data which must be handled by a server somewhere. Without the data centres containing these servers, the devices (or the apps they run) are of little value. Intel estimates that for every 400 smartphones, one new server is needed. That’s about one server per street I’d say.
We’re approaching 2 billion smartphones in service globally, each with (Intel estimates) an average of 26 apps installed. We check our phones an average of 110 times per day, and on top of that, each app needs to connect to its data centre around 20 times daily for updates. All of this adds up to around one trillion data centre accesses every day. And that’s just for smartphones. Out-of-home IoT devices like wearable medical devices or factory sensors need even more server resource.
Sounds like a lot, right? Actually, if we were watching a movie about the Internet, it’d be an epic and we’d still just be in the opening credits. Only about 40 percent of the world’s population is connected today, so there’s a huge amount of story yet to tell as more and more people come to use, like and expect on-demand, online services. With use of these applications and websites set to go up, and connected devices expected to reach 50 billion by 2020, your data centre is a critically important piece of your business.
Here Comes the Hybrid Cloud
What fascinates me about all this is the impact it’s going to have on the data centre and how we manage it. Businesses are finding that staggering volumes of data and demand for more complex analytics mean that they must be more responsive than ever before. They need to boost their agility and, as always, keep costs down – all in the face of this tsunami of connected devices and data.
The cost point is an important one. Its common knowledge that for a typical organisation, 75 percent of the IT budget goes on operating expenditure. In a bid to balance this cost/agility equation, many organizations have begun to adopt a Hybrid Cloud approach.
In the hybrid model, public cloud or SaaS is used to provide some of the more standard business services - such HR, expenses or CRM systems; but also to provide overspill capacity in times of peak demand. In turn, the private cloud hosts the organizations most sensitive or business-critical services, typically those delivering true business differentiating capabilities.
This hybrid cloud model may mean you get leading edge, regularly updated commodity services which consume less of your own valuable time and resource. However, to be truly effective your private cloud also needs to deliver highly efficient cost/agility dynamics – especially when faced with the dawning of the IoT age and its associated demands.
For many organizations the evolution of their data centre(s) to deliver upon the promise of private cloud is a journey they’ve been on for a number years, but one that’s brought near term benefits on the way. In fact, each stage in the journey should help drive time, cost and labour out of running your data centre.
The typical journey can be viewed as a series of milestones:
- Stage 1: Standardization. Consolidating storage, networking and compute resources across your data centres can create a simplified infrastructure that delivers cost and time savings. With standardized operating system, management tools and development platform, you can reduce the tools, skills, licensing and maintenance needed to run your IT.
- Stage 2: Virtualization. By virtualising your environment, you enable optimal use of compute resources, cutting the time needed to build new environments and eliminating the need to buy and operate one whole server for each application.
- Stage 3: Automation. Automated management of workloads and compute resource pools increases your data centre agility and helps save time. With real-time environment monitoring and automated provisioning and patching, you can do more with less.
- Stage 4: Orchestration. Highly agile, policy-based rapid and intelligent management of cloud resource pools can be achieved with full virtualization of compute, storage and networking into software-defined resource pools. This frees up your staff to focus on higher-value, non-routine assignments.
- Stage 5: Real-time Enterprise. Your ultra-agile, highly optimized, real-time management of federated cloud resources enables you to meet business-defined SLAs while monitoring your public and private cloud resources in real time. Fully automated management and composable resources enable your IT talent to focus on strategic imperatives for the business.
A typical reaction from organizations first considering the journey is “That sounds great!” However, this is quickly followed by two questions, the first being “Where do I begin?”
Well, let’s start with the fact that it’s hard to build a highly efficient cloud platform that will enable real-time decision making using old infrastructure. The hardware really does matter, and it needs to be modern, efficient and regularly refreshed – evergreen, if you will. If you don’t do this, you could be losing an awful lot of efficiency.
Did you know, for example, that according to a survey conducted by a Global 100 company in 2012, 32 percent of its servers were more than four years old? These servers made up just four percent of total server performance capabilities but yet they constituted 65 percent of the total energy consumption. Clearly, there are better ways to run a data centre.
It’s All About Meeting Business Expectations
And as for that second question? You guessed it, “How can we achieve steps 4 and 5?” This is a very real consideration, even for the most innovative of organisations. Even those companies considered leaders in their private cloud build-out are generally only at Stage 3: Automation, and experimenting with how to tackle Stage 4: Orchestration.
The key thing to remember is that your on-line services, web sites and apps run the show. They are a main point of contact with your customers (both internally and externally), so they must run smoothly and expectations must be met. This means your private cloud must be elastic – flexing on-demand as the businesses requires. Responding to business needs in weeks to months is no longer acceptable as the clock speed of business continues to ramp. Hours to minutes to seconds is the new order.
Time for a New Data Centre Architecture
I believe the best way to achieve this hyper-efficient yet agile private cloud model is to shift from the hardware-defined data centre of today to a new paradigm that is defined by the software: the software-defined infrastructure (SDI).
Does this mean I’m saying the infrastructure doesn’t matter? Not at all, and we’ll come on to this later in this blog series. I’ll be delving into the SDI architecture model in more detail, looking at what it is, Intel’s role in making it possible, and how it’ll enable your private cloud to get the Holy Grail – Stage 5: Real-time Enterprise.
In the meantime, I’d love to hear from you. How is your organization responding to the connected device deluge, and what does your journey to the hybrid cloud look like?