The hybrid cloud has arrived for the enterprise, but it comes with a complication: the speed of light. Between the cloud and the end user (including IoT devices that count as ‘users’), there is an emerging need for an intermediate environment that can satisfy real-time compute requirements without incurring the latency of reaching all the way to the cloud. ‘Edge compute’ is the phrase used to describe what covers this middle ground, but the optimal location for edge compute resources remains open to question.
As it turns out, there is more than one edge. There is certainly the far edge that reaches all the way to end users, mobile devices and remote sensors. But there is also a strong case for a near edge – an intermediate location that is closer than the cloud but doesn’t reach the granularity of individual IoT devices.
The edge is increasingly important because compute resources have become widely distributed. In the case of the public cloud, those resources can be in indeterminate faraway locations. At the same time, the sources of data have scattered outward. More enterprises are operating with distributed workforces, and the Internet of Things has crept into the enterprise lexicon as well.
Where should all that data be processed and stored, then? The immediate choices would seem to boil down to the user environment (out in the field, on the factory floor, inside a medical facility or bank) or the cloud. The former option – i.e., the far edge – is the suitable and even necessary choice for some verticals, but many enterprises will find it has impractical implications in terms of cost and complexity, especially where IoT devices are involved. Transporting that data to the cloud, however, takes time, and in an accelerating world where artificial intelligence and closed-loop automation promise new efficiencies, that latency represents a significant impediment. In some cases, it could be enough to break an application or ruin user experience.
This situation creates a strong case for the near edge. This environment would need to be robust enough to house compute and storage infrastructure and would also need the security and reliability that enterprises have come to expect from the cloud – and, of course, it would have to satisfy regulatory compliance requirements. While many near-edge options exist, one that seems promising for most enterprises is a multitenant datacenter, where network-enabled colocation would let the enterprise set up shop and connect easily to multiple clouds, SaaS resources and even partners and suppliers.