Automation for Physical Devices: the Holy Grail of Service Provisioning

Tuesday February 13, 2018

Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) are finally starting to pick up momentum. In the process, it is becoming clear that they are not the silver bullet originally advertised to be.

While great for some use cases, emerging technologies like SDN and NFV have been primarily designed for virtual greenfield environments. Yet large service providers continue to run tons of physical network devices that are still managed manually.

Based on discussions with senior executives at various service providers, the industry is gearing towards service agility and minimizing Operating Expenses (OPEX) through automation. But as fully automated workflows typically involve also physical network devices at select phases of the process, most network infrastructure vendors have been unable to go the whole nine yards together with their clients.

One of the obvious reasons why carriers have been hesitant to embrace automation is that any automated process is only as strong as its weakest link. By having to resort to manual steps towards the end of the process, the service agility suffers.

But perhaps even more importantly, partial automation abilities will diminish OPEX savings and limit the number of possible business cases. This is why automation for physical network devices is becoming the holy grail of service provisioning.

Enter Ansible – the Network Robot

Traditional orchestrators such as Chef, Puppet and Jenkins require physical agents to be installed on the managed devices. For large service providers with tens of thousands of devices to manage, this model is simply not practical. But over the last six months, the traditional approach has started giving way to agentless orchestration based on standard protocols such as SSH and SCP.

Pioneered by Red Hat with its Ansible network module, service providers are now able to weave the management of physical devices into their lifecycle orchestration models. For practical purposes, this is almost like placing a robot onto a network technician's seat, ensuring that changes to physical network devices are carried out automatically.

Because Ansible is an open source solution backed up by nearly every major vendor in the industry, the breadth of the ecosystem also enables valuable multi-vendor scenarios. This is important because it allows automated processes to run all the way from cloud portals to the physical devices on the ground. Given some time, this will be nothing less than revolutionary in unleashing the digital transformation.

Spreadsheets that Choked the Robot

The curious thing about network management is that there are typically no sophisticated solutions in place for managing VLAN spaces, Virtual Routing Functions (VRFs) and their connections with logical networks. Instead, the most common tool used for this purpose is a humble spreadsheet.

Considering that automated management of physical network devices relies heavily on assigning suitable VLANs, networks, and device-specific configuration parameters, the last manual hurdle for automated network services is the spreadsheet used to manage them. Without a backend from which to query all these properties, initiatives aimed at end-to-end automation are likely to hit a wall.

To eliminate the spreadsheets that choke the network robots, orchestrators need a single backend they can use to obtain all network-related data needed to configure devices. Here is a simple three-step methodology for unleashing the network robot:

1) Merge the entire network structure including logical networks, VLAN spaces and VRFs into a unified management system. This backend should provide all orchestrators with a simple REST-based API from which they can query free network resources and device-specific configurations automatically.

2) To ensure smooth end-to-end automation across various operational silos, make sure that the unified management system has a flexible data structure that accommodates different automation use cases. In the future, service automation is likely to give birth to entirely new network services, so having a high level of adaptability will be the key in enabling entirely new use cases.

3) Whenever automated changes are effected in the network, the unified management system should provide a single source of truth for all network information ensuring visibility, audit trails and compliance. This is an important consideration because otherwise, the engineers will lose the visibility they have come to enjoy while tasks have been carried out manually.

Although it has taken more than half a decade to reach this point, it now appears that emerging technologies are becoming mature enough to take a leapfrog into digital transformation. Looking into the 2020s, this will not only change the dynamics of the service provisioning industry, but also power innovation among the companies that leverage the next generation of digital platforms implemented today.



By Juha Holkkola, Co-Founder and Chief Technologist at FusionLayer Inc.Juha Holkkola is the Co-Founder and Chief Technologist at FusionLayer Inc. An inventor with several patents in the US and Europe, he is an advocate of technology concepts with tangible operational impact. Juha is an active proponent of emerging technology trends such as cloud computing, hybrid IT and network functions virtualization, and a regular speaker at various industry events.