Saturday, November 25, 2017

‘Hard and difficult is what we do,’ EdgeConneX chief on the emerging edge computing market

The opportunity across the edge computing market consequently brings with it the necessity to work hard to differentiate a business from another… business.

EdgeConneX has been around since 2013 and has delivered dozens of data centres with several more scheduled to come online in the near future. The company’s Chief Architect & VP of Innovation, Phill Lawson-Shanks, talks to João Marques Lima about the market and the business.


How do you differentiate fog computing from edge computing?

EdgeConneX’s Chief Architect & VP of Innovation, Phill Lawson-Shanks

PLS: Fog Computing was first explained in a paper by Cisco to describe the future of highly distributed technologies at the very periphery of the network – like IoT.

We believe that services need to reside and be transacted at or delivered from the location that best suits the consumer of these services.

In many cases this requires infrastructure and technological platforms to be positioned in locations at the lowest possible latency to the consumer – dependent upon the method and context of their connection to it.

If via broadband, then as close as possible to the Headend node, if over wireless, then at the cell tower, small cell or in-building distributed array. This is what we define as Edge computing.


Where does the barrier lie in terms of defining an edge data centre and a central computing node?

PLS: Edge is defined by its location in relation to the Internet’s core interconnection locations – like Ashburn VA, London, Paris and Amsterdam.

As in – it is not at the networks core but enables Internet Exchange Providers (IXP’s) to create and establish new places for traffic to peer and be delivered from. The Edge is first and foremost network-centric.

Simply building a data centre in a second tier market, but forcing the traffic to trombone back and forth from the networks core can never be called an Edge.

Having said that, as the central compute nodes have grown in capacity – and more importantly, server density – they have migrated away from the traditional data centre campus model that sprung up around the network core moving towards large mega-cloud sites where land and (more importantly) power is cheap(er).

However, as more and more workloads have moved to these mega-clouds, latency has started to become an increasing performance limitation.

To that end, we are now building a new breed of cloud node data centres at new core and edge locations for the hyperscale providers – to bring their services as close as possible to their on-net customers.


What is the hardest thing for a company like yours to deliver in the edge arena?

This article originally appeared in the last issue of the Data Economy Magazine. Click here to read more.

PLS: We have built our business on challenging and overturning what the market has thought of as difficult or impossible. Four years ago, we coined the term ‘Edge Data Centers’ and set out to define a new market “The Edge” – hard and difficult is what we do – in order to build out an infrastructure for our clients to deliver the next generation of internet-based services from.

EdgeConnex currently has a lot of DCs either under construction or being considered.


What are the plans for these?

PLS: We are constantly assessing the needs of our current and future clients to identify new locations that are or will become vital to their continued success.

It is not our business model to build speculatively – but only with an anchor customer where we can develop and facilitate the growth of an ecosystem.

This is made possible by our rapid build process and highly capable engineering and program management teams.

Our typical build process allows our customer to be installed and service–ready within six months signing a contract and agreeing on the site location.

We built our first 23 data centres within our first 24 months and now four and a half years later, have 31 active locations with another 10 in construction.

Due to the nature of our clients, we do not widely publicise our development plans, however, with 9 campuses and over 550WM delivered in 5 years of operation – I would say our model is finding strong market acceptance.


Interestingly, EdgeConnex has no data centres planned for Asia. Why’s that?

PLS: Our customers continue to ask and we continue to investigate Asia. At this time, however, we have nothing to report on future markets in Asia.


Beyond edge, what other trends do you see that are set to disrupt the data centre space as we know it?

PLS: Everyone is talking about autonomous vehicles, industrial and commercial IoT, VR, AR, and MR, etc. All of these will demand higher levels of compute, storage and network as close as possible to the device, so all present enormous areas for growth – at the edge and the core.

I do think that distributed ledgers (like blockchain and Ethereum) will drive incredible changes in how we deliver certified data, content, code around the internet.

Along with the emerging AI as service platforms next-generation Internet Exchanges and dynamic workflow control/distribution via intelligent orchestration engines will evolve rapidly.