How Closed-Loop Automation Advances Network Automation
Today’s applications and workloads move and change, are created and destroyed, faster than IT operations can keep up. Cloud bursting results in workloads moving automatically from on-premises infrastructure to public cloud resources. The proliferation of overlays adds network complexity. And although IT is more dynamic than ever, many network operations (NetOps) teams still work in the same, traditional way.
Most NetOps teams manage network devices one at a time. This process is a tedious, inefficient and error-prone way to support large numbers of switches, routers, firewalls, load balancers and so on. In even modest-sized networks, when teams deploy and manage a new application, they have to touch dozens, hundreds or even thousands of devices. And if the NetOps team is understaffed and overwhelmed, what should be a simple matter of copying and pasting a new configuration can turn into a service delivery nightmare.
Even though automation has moved these teams slightly forward, the collection of scripts they’re using to handle device configuration does nothing to continually validate the network in real time and identify problems as they occur. Today’s NetOps teams need a better approach to meet increasing business needs.
Evolving Network Automation
Closed-loop automation is evolving as the next major advancement in network automation. It’s the process of:
- Programmatically and continually validating the network.
- Automatically remediating problems.
- Validating the network again after remediation changes are pushed.
A closed-loop automation platform uses scripts and orchestration workflows to achieve optimal efficiency. But ultimately, it’s much more than an advanced orchestration tool.
Closed-loop automation allows a NetOps team to actively manage a large network with minimal staff and fewer errors, resulting in faster deployment times and quicker remediation.
Rather than yet another network management platform, closed-loop automation is a way to manage an entire infrastructure as a single system, increasing both efficiency and consistency — both of which are sorely lacking in NetOps today.
Closed loop automation directly impacts NetOps by:
- Decreasing human intervention for common troubleshooting tasks.
- Decreasing the mean time to resolution for incidents.
- Making network configuration across devices more consistent.
- Decreasing the inefficiency and inaccuracy of managing devices one at a time.
These measurable improvements increase network reliability and uptime. They also decrease the overall cost of network operations.
For example, rolling out a new application requires touching many — if not all — devices in a network. Specifically, deploying a new inventory management application may require new virtual servers, new VLANs, new routing policies, new quality-of-service (QoS) policies, new DNS entries and new firewall rules.
If this application will be used by wireless tablets on a factory floor, then the wireless LAN controllers (along with every single switch and router in the path between the tablet and the servers) must have the appropriate QoS policies configured correctly. This requires skill, accuracy, consistency and a significant amount of time. If any one of those components is missing, network operations can’t deliver.
Using a closed-loop automation platform automates this process on Day 0. It also validates and maintains the network on Day 2. This is the real benefit of closed-loop automation. It’s a continual cycle of programmatic discovery, validation and configuration — all tasks normally done by an entire NetOps team.
A Paradigm Shift in NetOps
Implementing closed-loop automation can be difficult for IT teams. It requires a certain level of trust between network operators and the layer of software that sits between them and their network devices. Trusting an advanced platform to automatically remediate issues may be a difficult mindset change for some IT managers. But, consider that dynamic routing protocols already do this as a matter of their inherent functionality.
When a WAN link goes down, for example, a typical routing protocol has enough intelligence built in to switch to a secondary path and maintain connectivity. In fact, organizations put their entire business on the line every day trusting that their preferred routing protocol will automatically remediate link failures.
In the compute world, VMware vCloud NFV, VMware vRealize Orchestrator and VMware vRealize Operations Manager work in concert to monitor and validate a virtual environment against a pre-determined reference architecture — or in other words, the intent of the network operator. In this way, when an issue is discovered in the virtual ecosystem, an alert can be generated. Then, an automatic remediation workflow can be performed by the system autonomously.
In the networking world, closed-loop automation platforms:
- Monitor network resources in real time.
- Validate that the network state conforms to the intent of the operator.
- Identify and isolate issues.
And as network operators trust these platforms more, the dream of autonomous remediation closes the automation loop.
It isn’t necessary to rip and replace an entire infrastructure or take an entire network offline to take advantage of closed-loop automation. Because these platforms interact with the network without being part of the data plane, organizations can deploy in a brownfield environment and in small doses.
At the outset, a closed-loop automation platform can be used to discover, monitor and manage a main distribution frame (MDF) or a secondary data center. In this way, network operations can begin to realize some of the benefits without handing over the keys to the entire network. The expense of shifting network operations workflows can be reduced and stretched over a longer operational cycle.
Closed-loop automation is one of the biggest trends in infrastructure management directly impacting the agility and efficiency of network operations.
Ultimately, closed-loop automation is dynamic and adaptive — responding to the network in real time. This is a huge advancement over traditional automation and orchestration approaches and the future of cloud networking.