The Road to Network Configuration Automation

By Paresh Jain, Head of OSS Platform, Tata Consultancy Services

The fast growth in the telecom industry, not only in size and value, but also in the types and volume of traffic that goes through the networks and the complexity of the systems, has made network configuration ever more demanding on operators, because it holds the key to the quality of the network, how it is governed and operated. This is made more critical when 5G is increasingly becoming a reality. Much promise has been made about 5G’s lead use cases like enhanced mobile broadband, massive IoT, mission critical communication, etc. However, they also raise unprecedented demands on network properties, including high demand for edge computing, extremely high traffic volume, and extremely low latency. Further SDN/NFV inherently have high network configuration automation and management and demands the same from physical networks. This would need consolidation of physical network configuration and support NFV orchestration with its automation requirements.

Long gone are the days when a group of engineers sat around a table, pen in hand, agreeing on the configuration parameters, then writing a few command lines. This would work in a static network when configuration was done once and then the network was left run by these rules for a long time. With the fast-moving dynamic network of today and especially of tomorrow, these key shortcomings of manual configuration management are evidently not able to cope with the network complexity:

• It is time consuming: dynamic networks need response to issues in minutes, not hours;
• It is prone to errors: the more complex the networks get, the more likely some aspects get overlooked, e.g. some system vulnerability that may be open to attacks, or subprime performance of some network components;
• This is a piecemeal approach, attempted to configure bit by bit, while the dynamic networks need a wholesale solution, from device management, data collection and processing, to trouble detection and problem solving;
• It cannot scale or customise, as every new network setup, e.g. a new vendor’s equipment integrated, needs a new set of command scripts.

In contrast, the automated configuration management solutions can:

• Save time and cost: this is especially in cases of configuring and managing mash-up cross-generation networks supplied by multiple vendors. However, this does not mean companies do not have to invest. Long term cost, especially OPEX, savings can only be realised if service providers are ready to invest in leading automated configuration systems;
• Minimise human errors: this will become even more apparent in 5G era when the complexity will be added with slicing of network, and for most operators it will be a co-existed mix with legacy networks for many years;
• Detect and solve problems more quickly: using audit function can pro-actively detect abnormal network behaviours almost real-time, and alarm the system and respond in the amount of time it takes the system to run the automated commands;
• Easily customise: by definition, automated configuration is done by software, and can be customised based on the use case and network context.

However, in order to properly implement an automated configuration, a step-by-step approach needs to be adopted. To start with, the automation set-up needs to have an automated system backup mechanism in place, whereby provisions are defined and maintained to regularly back up all the critical parameters for system elements as well as perform version control. The set-up should also be able to alert the management system when a backup fails to initiate or complete.

The next step to implement automated configuration is to identify network parameters and analyse how they affect the security, performance, availability, and other network service factors. All these parameters and their interactions with the network components should be documented for input to the configuration solution.
The most critical step in implementing network configuration automation is to automate network auditing. The auditing system should be capable of:

• Regularly monitoring network elements;
• Estimating and warning impact on service parameters impacted by configuration changes;
• Grouping audit rules to categories to cater for increasing demands;
• Taking remedial actions when system breach happens including rolling back to the last secure configuration;
• Generating detailed breach reports

To complete the cycle of network configuration automation, the management system should also be capable of pushing out automated configuration to all the devices in the network. This is particularly meaningful when large numbers of devices are deployed in a network that all need to meet the same level of security and function requirements, for example professional mobile service, or industrial IoT networks. Again, the roll-back capability is critical so that, in case configuration is not completed on certain devices, they can be automatically rolled back to the previous functional version, instead of being left in a limbo.

The industry readiness of automated network configuration has gone way beyond blueprint on paper, and it is encouraging to see strong interest from both the supply and demand sides. On the supply side, leading technology vendors already have commercial solutions to offer, for instance the HOBS Connected Devices Management solution from TCS. The solution covers the configuration management for Telco’s as well as Enterprise networks. It has shown considerable network quality and performance improvements within months of deployment in leading Telco networks. On the demand side, more and more service providers are embracing automated configuration management, for example increasingly RFIs and RFQs are including automated configuration in their requirements. There is no surprise in the increasing enthusiasm though as we have observed that automation will enable faster deployment of new services, therefore faster time to market, and faster return on investment.