LAS VEGAS -- Faster application development, Internet-enabled everything, big data analytics and other IT drivers are reshaping the data center infrastructure architecture that supports it all.
Data centers are growing up into cloud infrastructures and vice versa. Software is redefining hardware. While cloud use is growing and shaping user expectations, not every workload belongs in the cloud, said Henrique Cecci, research director at Gartner Inc., an IT research firm based in Stamford, Conn. Here's what he told SearchDataCenter about cloud and data center growth during the analyst firm's Data Center Conference in December 2013.
What's driving IT demands?
Henrique Cecci: Consider the trend of application acceleration. Mobile applications are changing the way we deal with applications; you're used to going to a portal to download an app. This changed the way we see and deploy applications and how they affect the data center. We're deploying more applications faster than before. In the past, this was not that simple because of testing, etc.
Applications advance infrastructure on the other side. If you're deploying applications faster, it requires faster deployments of hardware and infrastructure to support this new, faster world of applications. If I have the capability to deploy more applications, and faster, I have to provide back-end infrastructure architecture to deliver what's required.
'Internet of Things' also demands more IT infrastructure. It's getting bigger and bigger and generating a lot of data, a lot of storage and a lot of processing. If you double the size of the Internet of Things, you double your requirements.
How are data centers modernizing to meet demand?
Cecci: Let's talk about servers. There is 10% year-over-year growth on average. Storage is 50% growth; networking is 25% growth. Energy consumption is up 10% year over year, on average. We have lots of demands for growth in terms of energy, servers, networking -- the demand for IT is big. Data center managers have to deal with these new demands for growth. How is my capacity planning? How is my data center limited in terms of growth? How many years will I support these new demands using my existing infrastructure?
Cloud is not for everything; cloud is for specific kinds of workloads that make sense.
research director, Gartner Inc.
Software-defined everything -- it's one trend. In terms of cost efficiency, you can manage your workloads in a simple, less expensive way because you don't need to physically manage it; you can manage with software, not hardware. This is changing how we manage the network. It costs a lot to manage the network, so actually, the existing network architecture could be a problem for cloud computing growth.
You'll find integrated systems from Oracle with Exalogic, VCE with vBlock; NetApp [FlexPod] and other solutions. You see this from IBM, HP, all major vendors. [The companies buying converged infrastructure are] looking for higher performance. … CI is getting adopted more and more ... across all verticals. Enterprises won't use it for everything, but for specific workloads it makes sense.
Then there's the Open Compute Project. These guys are changing the way we design racks, servers [and] workloads. It's an important initiative. In three to five years, it will trickle down to medium and large enterprises. It's an open source community; you can download specifications for racks, for servers, for everything, and the goal is to produce very low-cost solutions. Companies are learning how to do these things from Open Compute Project initiatives, because all the information is readily available and really good.
What are the trends in hybrid cloud? Are data centers involved in the cloud and pulling workloads between data center servers and the cloud?
Cecci: The growth in hybrid cloud services and hybrid IT means enterprises are uploading part of their workloads to cloud -- public or private -- and dealing with more complex problems with several vendors.
Cloud is not for everything; cloud is for specific kinds of workloads that make sense. Usually not core applications, but maybe testing and development or [human resources]. What application acceleration does is drive more virtualization. It drives infrastructure that is able to scale.
What is changing in technology refresh cycles and capacity planning?
The POV from Gartner Data Center Conference
Five hyperscale IT methodologies to follow
Clouds tied up by automation glut
Cecci: [Data center managers] do the math; how much money am I saving on energy by replacing old stuff with this new equipment that's much more efficient? Let's suppose my [data center equipment] lifecycle is three years. To run a server for three years -- in some countries it costs more in energy than the price of the server. A new generation will save 20% to 30% energy, so it's about reducing the lifecycle to not only cost less, but process more data.
What about refreshing the software: virtualization tools and workload management tools that run on top of hardware?
Cecci: It's like a puzzle. You have to put all these pieces together and make your best decision possible. For example, we have DCIM [data center infrastructure management] tools. You can use results to compare vendors. But software is becoming more and more important.
Not many IT organizations are paying attention to energy. They don't have accountability to pay energy costs. Usually there is a facility guy responsible for that bill. The responsibility to update software doesn't fall on the facility guy. But IT is hugely responsible for energy use, and IT is realizing that energy is a valuable asset and something we should pay attention to. All the vendors have an environmental approach or green solution or DCIM solution to deploy.
The point is to focus on the total cost. If you are reducing the temperature in the data center by 1 degree Celcius, that may represent 4% energy costs in the data center. If you have the right temperature, you may save a lot of money.
This was first published in January 2014