dcsimg
 

The Networking Bots Are Coming

Wednesday Dec 27th 2017 by Arthur Cole

Automation could simplify many networking tasks — but it also poses new risks for massive errors that replicate at scale.

The march toward greater automation of network infrastructure will likely continue unabated in the new year — although it might not be as easy, or as straightforward, as it seems.

New platforms are rolling out at a steady clip. They promise the eventual transition from simple automation to intelligence-driven autonomy, but this should not blind networking executives to the fact that these technologies must be deployed carefully with safeguards in place to prevent potentially catastrophic disruption to data operations.

The next major advance to hit the channel will likely be the new bots that Juniper has added to its Contrail and AppFormix platforms. Together, these promise to fulfill the company’s vision of the intent-based, self-driving network in which human operators merely define the outcomes they require and the bots do the rest. The package consists of a peering bot that oversees things like Border Gateway Protocol (BGP) routing and policy enforcement, as well as a test bot for DevOps and CI/CD environments and a health bot for real-time data collection and analysis. In addition, the company has pushed the API framework of the Juniper Extension Toolkit (JET) to the network data plane, giving applications direct access to the company’s MX and vMX edge routers.

But despite this steadily advancing technology, don’t expect automation to be the cure-all for enterprise networking woes. Juniper’s own James Kelly uses an old adage to point out one potential flaw: “To err is human; to propagate errors massively at scale is automation.” In other words, when mistakes are built into network processes to begin with, they can lead to massive failures once the repetitive nature of automation kicks in. One of the biggest mistakes network engineers make is to identify a key problem and then attempt to hack it with automation. This is a sure way to produce conflict and confusion with the broader data environment since it is likely that other engineers are doing the same thing to solve their problems. The smart approach is to step back from the day-to-day and develop automation in a strategic fashion, targeting enterprise-wide objectives. Once that is in place, automating specific tasks becomes easier because it takes place within a properly defined framework.

Putting autonomous bots in charge of network provisioning and management can also lead to unintended complications if they are not subject to adequate auditing and compliance. Intelligent bot developer Talla discovered this after some 2,500 deployments at numerous companies. It created the BotChain network that collects data on bot activities to ensure they are behaving according to the rules. The system creates a digital certificate for every bot action and then stores the certificates in a blockchain where they can be assessed and analyzed. Not only does this provide crucial support for highly regulated environments, such as healthcare and finance, but it also provides a means to ensure that your bots have not been hacked by outsiders.

Bots have already infiltrated social media, ecommerce and a host of other digital functions, so there is no reason to suspect they will stay clear of the enterprise network much longer. But as with most technologies, success is usually a matter of proper execution, which in turn requires a clear understanding of the goals and objectives to be met.

Enterprise infrastructure has reached a point where virtually every upgrade must be made as part of a holistic, strategic vision. The last thing any organization needs is an army of disjointed, uncoordinated bots running amok with the keys to network infrastructure.

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