×
We have made updates to our Privacy Policy to reflect the implementation of the General Data Protection Regulation.
From Automation to Autonomy to… ?
 

From Automation to Autonomy to… ?

Thursday Jun 28th 2018 by Arthur Cole

Networks may one day become more adaptive and autonomous, but they should never be fully self-governing.

Network automation is the next phase in enterprise connectivity, but the real prize for many is the fully autonomous network. In this scenario, the operator merely defines what needs to be accomplished and the network, or the applications using it, define and provision the proper resources to get the job done.

But the enterprise has a long way to go before it can leverage this level of functionality in a meaningful way. And it seems improbable at this point that such a network will emerge without a clearer definition, and perhaps some substantial regulation, as to how it is to be used and what measures should be taken to protect it.

HPE’s Aruba unit took the latest step toward fully autonomous network recently with the launch of NetInsight, an intelligent analytics tool designed to improve visibility into complex mobile and IoT networks. The system is aimed at harnessing the increasingly large and diverse data sets needed to optimize connectivity and performance in enterprise environments, helping organizations shift from the current break-fix model of network management to a more holistic, self-correcting approach. At the same time, the company is expanding its ArubaEdge Technology Partner Program to broaden the reach of third-party solutions into the platform.

Still, many people remain concerned about adding too much intelligence to the network for fear that it will cause humans to lose control. This is the main worry behind Juniper’s plan to unleash software bots into enterprise infrastructure, says the company’s David Mihelcic, given the reputation that the technology has gained as a tool for malware and data theft. But bots can be highly beneficial when it comes to automating tasks that slow down provisioning, configuration and other tasks, and they can go a long way toward establishing a fully self-driving network in which technical staff can shed the nitty-gritty of network management and focus more on strategic development. They can also make the network much more reliable by reducing the kinds of errors that bring traffic to a standstill.

Autonomy is a big change from today’s semi-automated networks, so it’s important to tread cautiously when unleashing new capabilities into production environments, says Adva Optical’s Ulrich Kohn. This is where groups like the Metro Ethernet Forum can be a big help by establishing frameworks to help guide development. Using the group’s lifecycle service orchestration (LSO) architecture, organizations can assess basic technologies, map them against real-world requirements, identify issues and develop solutions to address those issues. The group has already defined a cloud-centric LSO that defines a hierarchical set of horizontal and vertical interfaces and an intra-provider north-south interface to manage abstract networks and service lifecycles.

But even full network autonomy is not necessarily the end-game for all of this development. Ciena’s Joe Cumello argues that the true aim of scale, automation and intelligence should be an adaptive network that can dynamically adjust itself to rapidly evolving data requirements. If the history of evolution has demonstrated anything, it’s that it is not the biggest, strongest or even the smartest that survives in the long run, but the most adaptable. Such a change is likely to be disruptive, however, and will require further development of a number of emerging technologies, including analytics, software control and programmable infrastructure.

Whether the network becomes autonomous, adaptive or something else, the enterprise should take care that it does not evolve from a self-managing entity into a self-governing one. At the end of the day, technology is the tool not the master, regardless of how intelligent it is.

At the moment, manual control of the network is interfering with the business model, but manual oversight will always be necessary to ensure that the needs of the data ecosystem do not supersede the needs of the user.

Disclosure: I provide occasional content development services to HPE and Adva Optical.

Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.

Home
Mobile Site | Full Site
Copyright 2018 © QuinStreet Inc. All Rights Reserved