Do I want to replace the lock in my front door with a radio-controlled device linked to a telco-hosted app on my mobile? Will company fleet managers want a cloud service to automate and remotely manage driving and maintenance functions of company cars — or why not top up with one of Google’s self-driving vehicles? Well, with the imminent arrival of the Internet of Things (IoT) it’s all in the pipeline — offering convenience and resource optimization. But as these functions proliferate, will the core network performance stand up to expectations, and is real-time error correction good enough?
Real-time network management was one of the topics at this week’s 2014 C-Scape event in London, where Cisco reiterated its strong belief in the imminent arrival of the IoT and over ever-more efficient, all-IP networks. Together these drive exponential traffic growth. With this follows increasing reliance on these processes in many aspects of our everyday lives. So the ability to spot and rectify error conditions before they seriously degrade or alter intended traffic flows, will become a serious competitive differentiator — ultimately a question of life and death, if for example, alarms from a network-connected pacemaker doesn’t reach the defibrillator or the right medical personnel in time!
The analogy in the material world is in the aviation field. Here manufacturers and public authorities painstakingly develop huge maintenance programs to ensure that airplane parts get replaced before they break, pilots go through a long check list process before taking off, and air traffic controllers make sure that when a plane is cleared for take-off it also has a slot to land in at its intended destination. When errors do occur and accidents happen, a well-defined and thoroughly standardized forensics process identifies root accident causes, and advises the whole aviation industry and operators of similar aircrafts to undertake error-correcting measures.
Network error conditions or ’anomalies’ are context specific and may be caused by software glitches, failing hardware – or malware/network attacks (DDoS etc.) Whatever the cause, fixing the problem before the end-user or the application is affected, and advising other networks with similar configurations of issues that may affect them, will require more network integration and a deeper cross-industry approach to network and service delivery. Four major requirements present themselves:
- Sensors and switching fabrics able to monitor traffic flows end-to-end and contextually identify ‘abnormal’ network flows
- Big data toots to analyse ‘anomalies’ in the traffic and identify root cause and resolution
- An upstream and downstream propagation mechanism for error correction, malware eradication and stopping network based attacks
- A global information dissemination process that informs all relevant parties of the anomaly characteristics and how to handle them.
A crucial step involves standardization of information and analysis exchanges between different network hardware and software components. That requires open standards and heralds the end of proprietary protocols [such as] as we know them today! This will be painful for a wide range of communication vendors – Microsoft’s Skype protocol being a prime example.
Telcos are rising to the challenge acknowledging that having ten different best-of-class routing fabrics in their fixed and mobile core network is just too complex to manage real-time, let alone proactively. Next generation components must all adhere to the standards being hammered out in the new world of software defined networks (SDN). Most likely upcoming telco infrastructure investments will go to vendors that address data flows from data center to edge router.
All this puts Cisco in pole position with its breadth of hardware and software products and application centric architectures that gets them closer to proactive, end-to-end application performance management. It also forces competitors both in the infrastructure (Ericsson, Huawei, Juniper, Nokia and Alcatel Lucent) servers (IBM, HP, Dell, Fujitsu) and network managers (CA and BMC) to get their acts together. Purchase decisions will no longer be based on price or best-of-breed but on total performance outcomes.