[This blog is an abbreviated version of the article “It’s Not Magic, It’s Machine Learning”]
Artificial intelligence is making headlines every day. It’s rapidly becoming a disruptive technology, with applications in the home, transportation, medicine and many other fields. While we’re still only at the start of our AI journey, we can already see that the potential is virtually unlimited. As AI continues to gain momentum, we’re going to see a hyper-connected world, with smart applications talking to us, to each other, and to a vast array of devices. AI is going to become mission-critical, working hand-in-hand with humans to create a radically new future.
However, here’s the question: How are we going to keep this new world up and running? As intelligent applications and the Internet of Things continue their explosive growth, we’re already seeing an exponential increase in the number of things we need to manage. And, it’s not just about the number. There’s also an enormous increase in complexity. For example, traditional monolithic architectures are being replaced by granular microservices – turning a single application into a complex network of interacting components.
Here’s the good news. AI doesn’t just create complexity – it also helps us to manage that complexity. Event management is a perfect example. While yesterday’s event management systems provided basic filtering and correlation capabilities, they weren’t designed to cope with the scale and complexity of today’s IT environments. As a result, IT operations teams are drowning in event noise and spending enormous amounts of time diagnosing and resolving service issues.
AI changes that. Today’s intelligent event management systems use AI to make sense out of the increasing flood of event data. Using a combination of configured logic and machine learning, they turn millions of raw events into a few actionable, prioritized alerts, helping humans to resolve service issues quickly and accurately.
However, there’s a common misperception – promoted by some vendors – that AI is magic, capable of doing anything while being beyond human understanding. That’s just not true, but it’s convenient. By positioning AI as magic, they can make broad-reaching claims without having to explain how – or if – it actually works.
At Evanios, we beg to differ. Properly done, machine learning is transparent and understandable, and it solves real IT operations issues out of the box. That’s essential. Unless AI delivers real and quantifiable value, why would you invest in it? And, unless you can look behind the curtain, you’re being asked to take things on blind faith. No one trusts what they can’t understand and control, creating enormous adoption issues for black-box machine learning solutions.
So, what are the problems that intelligent event management platforms address out of the box? Here are the ones that we see at Evanios every day:
- Eliminating event noise, providing a much deeper level of noise reduction than legacy script-based approaches. This means that you have a clean signal, so you see what matters.
- Prioritizing work, scoring events based on their business and service impact. As a result, you focus on the important things first, rather than trying to guess which issues are most critical.
- Automated root cause analysis, generating a ranked list of likely causes based on infrastructure architecture, spatial and temporal correlation of events, and analysis of similar historical events.
- Predicting future service issues, learning and detecting event patterns that typically precede service outages and degradations. This means you can take action before customers are affected.
- Creating cross-stack visibility, breaking down technology silos and adding business context, so that you eliminate war room finger-pointing and accelerate service restoration.
- Automating remediation, learning how similar issues were remediated in the past, and then suggesting appropriate remedial actions – or even orchestrating the entire remediation process.
Here’s perhaps the best part. Some intelligent event management platforms can do this without you having to teach them. If you choose an event management platform that supports unsupervised learning – a major step forward from supervised approaches – you don’t need data scientists to continually train the system, since it adapts automatically to your evolving IT environment. And, with an unsupervised learning approach, you can look behind the curtain to see what the system is doing – and why – and even tune that behavior when needed.
Ultimately, machine learning is about meeting the demands of an increasingly intelligent, connected and complex IT environment, where business needs are driving a rapid increase in digitalization and innovation. It’s not magic – if you choose the right vendor, it’s a pragmatic and understandable way of solving today’s and tomorrow’s IT operations challenges.
To find out more about how machine learning can help your IT operations team to work smarter, faster and better, download the full article “It’s Not Magic. It’s Machine Learning.”