One of the best barometers for the preoccupations of the data centre industry is the conference programme at BroadGroup’s annual Datacloud Congress in Monaco and this year’s line-up is no exception. Energy always figures; no programme would be complete without a discussion on how to use less of it. However the focus has changed from M&E solutions to a more strategic consideration of the way that energy flows between data centres, comms networks and digital devices, and the potential role that the sector can play in stimulating renewable generation.
This reflects a more general move in subject matter from specific technologies, applications and services to a broader discussion about much more seismic changes in the industry: so there is less talk about adiabatic lapse rates and computational fluid dynamics and more talk about Edge, about 5G, about hyperscale investment.
I think this is welcome: comparing widgets is a lot less interesting than thinking about how they might influence change within the sector. However, future gazing comes with risks: digital technologies are disruptive – that’s what it says on the tin – and as we know from past experience, anyone trying to predict how things will evolve may as well bark at the moon.
So with that in mind, let’s look at the most disruptive factor of all: connectivity. Connectivity controls the internet economy (which means pretty much all our digital activity) in two ways: speed and price. The faster the speed, the more opportunities there are for adopting digital media: video on demand only works if we aren’t waiting 45 minutes for it to download, or watching freeze frames. Digital communication is also very, very, very price elastic: the cheaper it is, the more we do. Compare this to a price inelastic commodity like a coffin – the total number of coffins sold does not depend on how much they cost. By contrast, think how we would use Facebook if there were a charge per image. The horror.
At Datacloud today, Edge is again taking centre stage, but we are also seeing increasing focus on 5G. This makes a lot of sense: in my view 5G and edge must and will co-evolve. 5G isn’t like previous Gs, it’ll be flexible enough to optimise for different use cases – and edge servers will be key to much of that.
While wireless connectivity for end users might seem to sit outside traditional dialogues about comms and network infrastructure, in reality the way we use personal devices profoundly impacts data centres. Workers are increasingly mobile, taking their office with them via portable devices and collaboration tools. 5G will accelerate that trend, increasing demand outside traditional peak times, and dispersing data demand (and need for edge servers) more widely.
We will cross a new frontier when we start moving from 4G to 5G from 2020. Connection and download speeds will rocket and capacity can be segmented for specific uses, continuing the trend to transmit larger and larger packets of data. But, critically, 5G is also optimised for small packets, enabling the internet of things, connecting autonomous vehicles with roadside infrastructure, making smart cities real.
On reflection, each iteration of communications infrastructure seems to usher in a new type of customer for digital services. Individual consumers were not significant customers of digital services until we had 4G. Now there is a whole suite of large providers dedicated to serving this relatively new market. 5G won’t just service that community, it will herald a completely new type of consumer: Things. It looks likely that this will augment, not displace, digital activity. By doing so 5G looks set to fuel the ongoing explosion in digital data.
Observers fear a parallel explosion in data centre numbers and consequent energy consumption but while we do need to keep a close eye on sector energy use, I think these claims are overstated. There are three things to bear in mind here before we all decide we have to go and live in caves and eat bugs because data centres have gobbled up all the power.
Firstly, data centre energy use is growing, but not in line with data volumes. Thanks to factors like Moore’s Law, virtualisation and cloud, the energy needed to process a given amount of data has decreased by over six orders of magnitude in 30 years.
Secondly, we need to consider what all these data-hungry smart things are doing: if they are improving efficiency, streamlining processes and enabling dematerialisation then they will deliver net reductions in energy use. M2M applications for autonomous vehicles, supported by 5G and delivered through edge data centres, should be capable of reducing congestion, optimising route choice, regulating traffic flows though pinch points, and minimising kind of stop-start driving that is so inefficient. Safety should also improve – that’s the theory, anyway
Thirdly, 5G is unlikely to drive an explosion in data centre energy demand because the business model depends on a very high level of activity at very low unit cost. The 5G network and any supporting edge infrastructure really do have to be super-efficient for IoT to work, and for services to be monetised.
Some observers consider that edge data centres have the potential to be far more autonomous in terms of electricity supply, and that a much wider range of power sources will be available to them, because they are less monolithic and resilience can be built in by duplication and overlap rather than by continuity of supply to individual units. So that could be another mitigating factor.
But we can’t be complacent. While I think that market forces will keep energy consumption under control, 5G and edge collectively represent significant growth in what is effectively a new form of distributed IT. During the last decade we have encouraged the consolidation of IT functions into larger, purpose-built facilities where electricity consumption is transparent, where energy stewardship is scrutinised and where there are strong incentives for efficiency. So my concern relates to how we aggregate that new energy use, how we ensure that it is transparent and accountable. How will we even know how much activity sits within this new-old business model? Much of my advocacy work on behalf of the sector boils down to arguing about how much energy we use, so I want to see more certainty, not less, around consumption.
That’s my view from the sidelines, but if you really want to know what’s going on in the world of 5G, tune in to Datacloud Global Congress where most of today’s programme is dedicated to exploring the implications of new technologies and infrastructures for the sector.