New data centre technologies required to support smart grids and self-driving cars

Intel and others are developing technologies to enable data centres to support the coming embedded computing revolution

A slew of new technologies set to either be released or to go mainstream over the next decade will transform the data centre, a transformation that will go hand-in-hand with a wider revolution in computing that will demand faster and more interactive data centres.

According to Richard Curran, director of enterprise marketing at Intel EMEA, an increasingly urbanised world will require quicker, more responsive and more interactive data centres to cope with the colossal volumes of data generated by "smart grids", "smart homes", sensors embedded almost everywhere and self-driving vehicles, among other developments.

Speaking at the Computing Data Centre Summit, Curran said: "We are looking at the capabilities of 'smart grids' and 'smart homes' and new services being available. Cars, intelligent cars and transportation generally will actually change and offer a much better service as a result of new intelligent capabilities in years to come."

New technology

Technologies that Intel is actively researching and expecting to be incorporated into devices in the foreseeable future include gesture recognition, voice, augmented reality, wireless charging and facial analysis, among other things, said Curran.

"Those devices will either be embedded or very much context aware, whether you're in a store or at home, and also collecting data and helping you and your suppliers give you a better service," he added.

The collection and manipulation of all that data will require major advances in data centre technology and capabilities. Indeed, some 90 per cent of the world's data was created in just the past two years - and this is just the start of the "data revolution".

The data centre requirements of academia, life sciences and financial services will grow exponentially as a result, claimed Curran, and other sectors will not be far behind. And all the data generated will need to be analysed and stored at some level.

For example, modern passenger aircraft generate 1TB of data from sensors every hour. "We're looking at the ability to collect the right kind of information and to send that information to a source that can analyse that information.

"That makes it possible to predict better quality, better service capabilities of the plane and also to predict where these planes need to be so that airlines can service them as quickly as possible because they only make money when they are in the air," said Curran.

In future, in the internet-of-things, it won't just be Airbus and Boeing planes that have such capabilities built in - it will be everything, and all the data generated will need to be rapidly analysed, categorised, stored and equally rapidly retrieved. That will require data centre capabilities exponentially greater than anything that exists today.

New data centre technologies required to support smart grids and self-driving cars

Intel and others are developing technologies to enable data centres to support the coming embedded computing revolution

Super computing

Even now, a one petaflop computer, which developers of high performance computers struggled to produce just five years ago, can be fitted in a rack, said Curran.

Indeed, the number 25 in the supercomputer "league table", he added, will soon be able to fit in a rack, while in the not too distant future it will be possible for a 2W mobile device to offer 100 gigaflops of performance. Curran also predicted the imminent arrival of a gigaflop of compute power in a 20 milliwatt embedded or wearable device.

"So the capabilities and performance of products that we will be able to bring to market in the next five or 10 years will be absolutely incredible," said Curran. "As a result, we really do believe that the internet-of-things is real and will takeoff. We are looking at energy, cities, agriculture, transportation and infrastructure as a whole."

They will be connected, intelligent and enable "end-to-end experiences", he added, which will therefore lead to a flood of data into the data centre in the near future.

"We look at various different workloads from input/output-intensive to compute-intensive capabilities... The workload topologies enable us to understand what kind of architecture we need to put in place," said Curran. "We've also looked at the software-defined networking capabilities that will be coming in; we've already announced capabilities that will enable people to take advantage of software-defined networks."

Service providers have embraced software-defined networking in their networks, in particular, as it can drive major savings in their network topologies, especially in association with their cloud-based and managed services.

In storage, Intel is looking at cache acceleration software alongside much faster - but more expensive - solid-state disks. "As more data comes into the data centre we need to understand it, not just from the server perspective, but also the network and storage, and how to manage it, including the fabric."

Intel is looking at the data centre as a system, at how the individual components must ultimately tie in together, said Curran. The server, for example, could be based on anything from current Intel Xeon microprocessors to, conceivably, photonics and switch fabrics; while storage could involve utilising solid-state disks and caching.

"Combining all those components together we are getting some incredible benefits from performance, cost savings and energy efficiency perspectives. We are working with a couple of global partners and looking to introduce some of these kinds of products soon," said Curran.

New data centre technologies required to support smart grids and self-driving cars

Intel and others are developing technologies to enable data centres to support the coming embedded computing revolution

Data centre expansion

Many companies, he continued, are tentatively weighing up expanding their data centres or moving to a hybrid cloud environment in which cloud provides an extension to an organisation's existing infrastructure.

"And then there's big data. We are not looking at it as a piece of technology. It's all about business and business transformation. It's about utilising and understanding the information. Data is king. It's got an incredible amount of value and it's about what you do with it that counts," said Curran.

But with embedded devices generating data and machine-to-machine communications, identifying which data is important will be challenging - as well as the cost of storing, analysing it and the sheer time it can take to process such data.

That's where solid-state drives with massive input/output capabilities come in, with the efficiencies that they provide in speed balancing their higher costs compared to conventional disk storage technology. Alternatively, they can be used as rapid-access caches holding the most frequently analysed data.

"We've optimised and tuned Hadoop to take full advantage of our own architecture, so it's much more performant than anything out there," said Curran. So, from taking as much as a day to process 1TB of data on a conventional database stored on mechanical disks, it's possible to take this down to seven minutes using a combination of Hadoop and solid-state disks, he added.

Curran suggested that the world is about to enter the "fourth industrial revolution" driven by a combination of the internet-of-things and big data - the internet-of-things generating the data, and big data techniques analysing it.

For organisations and people working in data centres in particular, "it's all about how you will involve yourself in business transformation in the forthcoming years" said Curran. "And how you can help your organisations and companies to transform. As a result, IT will be at the heart of that transformation... it is going to work and it is going to change things.

"You've got to get really close to line-of-business and understand where your decision makers want to take their services and capabilities, whether in HR, finance, manufacturing, R&D and the overall strategy of the organisation.

So you have to understand how it needs to change, evolve and modernise, not just in terms of its capabilities today, but where it's going to be in 2015, 2020? What are you going to do to create an environment that can service that?" asked Curran.