The current crisis has disrupted how businesses operate, forcing them to speed up digitization efforts to facilitate teleworking and adjust to staff fluctuations. AIOps has become essential for automating these processes -- now and in the future.
By abruptly forcing most people to work from home, and by triggering an economic crisis, the global pandemic has upended business operations.
Not only must business leaders facilitate remote work among their employees, but they must also accommodate new ways of interacting with suppliers, partners and customers. Meanwhile, businesses’ digital channels and infrastructure, already critical prior to the crisis, have become even more essential, and yet harder to monitor and manage.
In the middle of this maelstrom are IT Ops teams, which have been called to accelerate digitization efforts, improve telework systems, and prevent digital service disruptions and outages — all elements with a direct, concrete impact on business outcomes.
COVID-19 and Digital Business
As they step up to these challenges, many IT Ops teams are finding AIOps to be a key component of a successful strategy. Specifically, AIOps has become the foundation for IT Ops teams to build a virtual NOC (network operations center) that allows them, while working remotely, to effectively collaborate and communicate, and provide continuous assurance of key digital services that power their businesses.
That is the topic Moogsoft EMEA CTO and global AIOps expert Will Cappelli addressed during his webinar “Virtualize the NOC: Futureproof Your IT Investment with AIOps.” “I’ll discuss how the pandemic is changing how digital businesses are operating and how ITOps pros manage that digital business,” Cappelli said.
Listen to a recording of the webinar, where you’ll learn how AIOps:
- Reduces costs by automating IT operations and consolidating management tools.
- Rapidly detects unexpected system behavior from novel work patterns and system architecture strains and accelerates response time to problems
- Can advance the remote-work trend by facilitating smart collaboration with automated learning.
- Can minimize the pandemic’s unpredictable impact on workforce size, allowing fewer IT professionals to accomplish more meaningful tasks.
- Consists, when properly implemented, of five key algorithmic dimensions: data selection; pattern discovery; inference; collaboration; and automation.
Cappelli also answered questions from the audience, during and after the webcast. Below is an edited transcript of the Q&A.
Is AIOps a good way for an enterprise to get started in AI, or should an enterprise get some AI experience under its belt first?
There are a number of advantages to starting an enterprise’s overall AI efforts with AIOps. First, there’s a lot of AI already in traditional IT operations management tools, like those for event management, service desk, and performance management — they already use an older generation of AI. So IT Ops teams have experience with at least the basics of AI, whereas in other areas of the business, AI is a completely new experience.
Secondly, IT systems, even though they’re very complex, are ultimately trackable and deterministic. So it is easier to get good value results out of applying AI to ITOM issues, and applying AI to the analysis of IT system behavior, than it is to apply AI to the vagaries of business interactions.
So for an enterprise considering an AI journey, there’s a good case for taking the first step through AIOps.
Does AIOps complement or compete with Unified Monitoring products?
It’s definitely complementary to Unified Monitoring (UM). You want to get input from the performance monitoring systems for your applications, infrastructure, and networks, and then use AI not just to further synthesize those views but to analyze the more complex behaviors that the monitoring systems won’t capture. AIOps plays the further role of linking those UM systems to the array of service management and automation tools that you must bring to bear with regards to IT operations management.
So AIOps isn’t there to replace UM. It complements and enriches UM, and integrates it with the rest of IT operations management.
Are any dimensions of AIOps easier to start with than others, or should you do everything at once?
Although each of the five dimensions of AIOps adds a unique and tangible value, it’s very difficult to implement all five at once. I recommend starting in two areas.
One is data selection. You really need to cut through the noise, redundancy and high volume of data coming from complex systems before you can do anything else. Luckily, that doesn’t require much human intervention. You don’t need data scientists to get the most out of data selection technology because it’s largely “unsupervised” AI technology.
The other area is the collaboration dimension, which was important before the pandemic, and has now become essential. It’s very difficult to deliver NOC services with large teams scattered and working from home, and with the workforce fluctuating, unless you inject into that mix this AI that supports collaboration, which Moogsoft provides via the Situation Room.
You can use these two areas to start, and to put your stake in the ground, to then build up to the other three dimensions.
The Five Dimensions of AIOps
How can an organization decide what are the best use cases to tackle first as part of AIOps, and based on that where in the journey should they start engaging with Moogsoft?
You want to take your key business applications and score them in terms of their significance to the business. Then on another axis you want to measure the amount of self descriptive data these applications are generating, including logs, metrics, event records and so on. If you match that up, you’ll get good guidance about where you should start.
Remember that you’re looking for AIOps to help you support your digital business. It’s really about the economics of the situation. It’s not technology efficiency, although that’s a side effect. The key is to ensure that your most important business applications are servicing your customer community properly.
In terms of engaging Moogsoft, it ought to be engaged from the get go, precisely because our platform deals with all five dimensions of AIOps, and because it’s highly automated and modular. So if you have business applications that are critical and complex enough to require AIOps, as most large enterprises do, a platform like Moogsoft’s that deals with all five dimensions is something you want to start working with as soon as you get started on your AIOps journey.
The Moogsoft Approach to AIOps
Does it matter if the remote NOC workers are in the same continent or country. Any performance challenges? Are there many examples of connecting the US with Indian NOC/IT teams?
In theory, it should not matter. As is usually the case, there are often cultural differences that make communication difficult and they are almost never taken sufficiently into account. Having said that, if the virtual NOC is supported by AI technologies, particularly around the creation and accumulation of knowledge capital, the cultural barriers should be considerably easier to overcome. In effect, the AI will do a lot of the ‘heavy lifting’ when it comes to cultural translation.
What do you think about capacity planning as a use case?
Capacity planning is, indeed, an important use case. A virtual NOC infrastructure should make capacity planning easier, precisely because, if AI-enabled, it will be able to synthesize incoming and shifting requirements for capacity and calculate and recalculate the resources required when and where they are needed on a global basis. Everything I said about the complexity of incoming signals during the presentation also applies to complexity of determining resource requirements.
Are there some good open source implementations of AIOps technology?
There are many open source AI tool kits available and Moogsoft would make use of them, if we could. The problem, we have discovered, is that the mathematics required to perform analyses of rapidly changing data sets is not widely understood even in the academic community. So basically, our co-founder and CEO Phil Tee had to develop much of the core mathematics we use in our algorithms from the ground up.
If you enjoyed our Virtualize the NOC webinar series then join us for our weekly Lunch & Learn demo series, and learn how Moogsoft AIOps can help your Ops teams work remotely during this time of crisis — and in the future.