Computing on the Edge

Cloud is everywhere and in everything. From servers through applications and even into networking. Public, Private and Hybrid: there is a version for everybody, fitting every use case and every requirement. At least that's the theory, but over the last two years a new buzzword is threatening to dethrone the cloud from its hard won position at the centre of enterprise computing. That hot new trend is the edge.

The only problem with edge computing is that it's not new. In fact, most enterprises were doing it anyway to some degree or other and have been since their very beginning. At its most basic level, edge computing is little more hosting your own servers in a location close to the people using them. This is different from private cloud, which involves hosting servers in your own data center.

New Problems

The classic edge computing example is hosting a file server and domain controller in every branch office. In the age of slow or unreliable network connections this allowed every business location to continue working regardless of the status of the Internet connection to HQ. This concern no longer applies in most locations, but has been replaced by latency concerns in financial services and the requirements of IoT devices in manufacturing.

Internet downtime doesn't matter any more for general business use, given the flexibility of most office workers. It can, however, grind a manufacturing plant to a halt unless the servers that control the production line are in the same building as the production line itself.

In each of these scenarios, the physical distance between the user and the server is of crucial importance. Every additional mile or kilometre adds nanoseconds to the time between the user clicking a button and the server carrying out the requested action. If that action is a record being updated in the CRM system, then such small margins will make zero difference to business performance. If the action is a stock trade or other financial transaction then every nanosecond can make a critical difference to the bottom line, after price fluctuations and changing trading volumes are considered.

After the Cloud

The issues inherent in running data centres at scale have been solved by cloud computing. Therefore, the focus for IT departments and companies marketing to them has inevitably turned to deciding how to solve the same challenges for distributed devices and servers. The cloud after all requires a data center to be hosted in a small number of physical locations. Edge is the exact opposite.

As such, the challenge that the edge presents has required new technologies and new standards to be developed, particularly around security and availability. Failover still needs to be considered for edge servers even without the vast array of duplicate and triplicate standby hardware available in an enterprise scale cloud. Security is even harder because traditional network boundary based security schemes such as firewalls and IP whitelisting are no longer sufficient to keep hackers out.

Old Solutions

At the same time, many of the solutions being pitched as edge computing were also being pitched as cloud solutions a few years ago. That applies to server and networking vendors whose products can fit both cloud and edge use cases depending on the buyer. Co-location vendors are doing it too, which may sound odd given that by definition this involves placing your servers with a third party in their data centre - a common description for the public cloud. Yet many data centres run by Equinix and similar firms were deliberately built near major financial and industrial centres, precisely so they cloud attract the financial and industrial firms that also comprise the core audience for the edge computing market. With cloud now synonymous with the major PaaS and IaaS vendors, Edge is actually a great way to pitch co-location services.

Every important business or technology trend inevitably gets a lot of bandwagoning from marketers looking for a new angle to pitch old products. Witness the entire security industry trying to pitch firewalls and anti-virus software as a GDPR solution twelve months ago. There is nothing wrong with this, but an informed buyer will always cut past the hype. When considering the average enterprise purchase, the buyer almost always knows more about the market than the marketer trying to sell the product. They'll have done their research beforehand, so can eventually see through any attempt to pitch a product as something it isn't.