Your browser is not supported. Please update it.

6 December 2018

AWS goes on-premise with edge cloud, stealing a telco advantage

The second Re:invent announcement that may strike fear into operators – especially those placing high hopes on the edge computing trend to enhance their business models – is Outposts. This sees AWS, in partnership with VMware, getting into the on-premises hardware business.

Control of a physical industrial edge has been assumed to be a stretch for the centralized webscalers, and therefore an area where operators (telcos or enterprise network providers) could more easily seize the best place in the value chain. But Outposts is a way to bring the AWS cloud into the enterprise premises by offering enterprises local compute and storage, based on the same hardware as AWS uses in its own data centers.

Users can run VMware private cloud software on Outposts, or the native AWS APIs (application programming interfaces). Either route aims to create a uniform experience across the public and private cloud, and to help companies move more smoothly towards the former.

AWS CEO Andy Jassy said in his keynote at re:Invent that some companies need more help to move from private to public cloud. “The longer they wait, the deeper the ditch gets,” he said. One of the deterrents is that many industrial companies require low latency in their cloud applications and would rather keep them on-premise to improve response times as well as data security and control. Outposts allows these firms to run the compute and storage on their premises, while interacting seamlessly with AWS applications and services.

VMware introduced VMware Cloud to the AWS public cloud last year and will now extend it to Outposts, via the virtualization firm’s Software Defined Data Center (SDDC).

AWS Outposts are in private preview, with public general availability in the second half of next year.

This launch follows in the footsteps of AWS’s first edge hardware offering, Snowball Edge, which is an appliance designed mainly for rugged conditions such as factory floors. And Microsoft has taken a similar path with Azure Stack, which runs Azure infrastructure and platform software on on-premises hardware; while Google recently unveiled its GKE On-Prem Kubernetes server.


While these developments do not threaten telcos directly, they do see the webscalers expanding into the physical edge in a way that will leave little oxygen for other value chain partners to breathe. Combined with connectivity under the enterprise’s own control, the edge can support a self-contained cloud which can address rising concerns about security and privacy, and the location of data; as well as enable emerging processes and applications which require low latency, especially in the IoT. But to make sense for the telco, that connectivity needs to be more than a bitpipe to an AWS server.

In the main mobile network, the operator is well positioned to deploy edge cloud resources, perhaps at cell sites or in central offices, and these locations are adequate to support classic telco services like better quality video streaming or personalized marketing. As long as edge success is tied to physical locations, operators are better placed then webscalers – they have highly distributed networks and sites, while the cloud giants’ economics rely on centralization, not building out thousands of mini-data centers.

However, many enterprises want their edge clouds in their own premises, which robs the operator of its location advantage. A telco may still extend their enterprise connectivity offerings into the edge, but they have to compete for the business with companies coming from the cloud side of the equation. In this scenario, the webscaler may not only offer local hardware, but take on an orchestration and coordination role, using virtualized connectivity to tie together multiple edges and physical connections from different operators – squeezing them into the commodity role.