How We’re Using AI and ML to Improve Cloud Management
With artificial intelligence and machine learning in the news of late, what might they mean for cloud computing? Michael Vitale explains how we’re positioning these technologies to benefit our customers.
My retired father called me this week to ask if I’ve heard about ChatGPT. He was amazed at the capabilities of the application and asked me where this was headed. It was at this exact moment I knew artificial intelligence (AI) and machine learning (ML) were entering a new phase.
It wasn’t long ago that AI and ML were topics for science fiction, but in the first quarter of 2023, years of hard work and incremental breakthroughs have produced some of the most sophisticated software anyone has ever seen. Every major tech player is throwing their hat in the ring, from Facebook’s LLaMA to Google’s Bard looking to take on Open AI’s ChatGPT head-on.
AI and ML applications for the cloud
At Megaport, we believe that AI and ML will transform cloud networking. As our global customers experience staffing challenges with their network and cloud teams, we wanted to explore how AI and machine learning could automate tasks, such as forecasting critical resources, to help reduce costs and avert downtime.
Data is crucial to any ML-powered forecast. As an industry-leading Network as a Service (NaaS) provider, Megaport has access to large amounts of historical bandwidth utilization data. Our patent-pending solution uses ML algorithms to analyze past utilization to produce actionable forecasts for network capacity planning. If the application detects bandwidth usage exceeding pre-defined thresholds, clients are alerted weeks in advance.
In the example use case shown below, a customer is connecting a Megaport Cloud Router to an AWS cloud region. Shaded in blue is actual bandwidth, where an artificial cap of 100 Mbps has been set. Shaded in purple is a corresponding ML generated forecast for the circuit’s predicted bandwidth utilization over the next four weeks.
This visual represents a tremendous amount of historical data in an easily absorbed view. Customers can customize alerts to notify them if any bandwidth thresholds are predicted to be met, averting any potential downtime.
Proceed with caution
There’s always the potential for a powerful tool to be misused. Today’s AI performs strictly-defined tasks, such as natural language processing or internet searches. But in the future, a more generalized AI, where systems can handle any task that humans can (and probably better), has some people concerned, particularly in the following areas:
- Security – If AI is vulnerable to hacking, it could pose extreme risks when overseeing a critical system such as a power grid. You can learn how to bolster your network’s security in our blog.
- Social manipulation – Social media algorithms are very effective for target marketing. However, this targeting has also been used for spreading misinformation.
- Privacy – With cameras in many places, facial recognition makes it possible to identify individuals and track where they are, with huge implications for people’s privacy.
As ubiquitous as AI has become, people still lack confidence in its accuracy, and often with good reason. Because AI is often working behind the scenes, many people simply aren’t aware of all the times it works flawlessly. Instead, the moments it goes wrong leave a lasting impression. Those of us working with ML and AI have our work cut out for us if we want to deliver it in a manner that inspires trust in its users.
Building for the future
Today’s use of ML within Megaport’s platform still requires a human to respond to an alert and take action. In the near future we plan a full “autopilot” experience where an organization can set thresholds based on actual or predicted utilization, and then choose from one of three options:
- Send a message when the threshold is reached;
- Notify and schedule an upgrade in the next maintenance window, or;
- Automatically upgrade when the point is reached and notify.
This innovation puts data and control into the hands of our customers, enabling them to more efficiently manage their cloud infrastructure and adjust resourcing before it causes downtime, and ensuring that their network is continually evolving in lockstep with their usage.
At Megaport, we view AI and ML as important tools in our mission to be a trusted partner providing our customers with greater simplicity, choice, and control as they harness the power of the cloud, and to continuously evolve our platform in anticipation of where our customers’ network needs are headed.