What is AIOps? Are you confused by all the different explanations out there? Regardless of the mixed messages, there is a correct definition and approach for AIOps.
Has your IT environment become exponentially complex as it has been modernized? As a result, are you looking for more powerful IT monitoring and management technologies? If so, you’ve probably heard of AIOps.
Indeed, thousands of organizations globally have successfully leveraged AIOps to automate incident management, accelerate problem resolution and prevent service outages in their modern IT environments.
But as you look deeper into AIOps — the application of artificial intelligence to IT operations — you’ll realize that it isn’t a niftily shrinkwrapped “product.” There’s no consensus on which features define an AIOps solution. Instead, you’ll find various opinions from multiple vendors.
However, as a leader and pioneer in this space, Moogsoft believes there is a correct approach to AIOps. That’s what Will Cappelli, Moogsoft EMEA CTO and Gartner’s former leading AIOps analyst, explains in the webinar “How to Use the Power of AI to Transform IT Operations.”
Cappelli outlines the fundamentals of what defines AIOps, and helps you get clarity on how to evaluate market offerings. For example, IT operations management vendors are adding AI features to their products, but AIOps platforms deliver the most value by coordinating those domain-specific tools, he says.
He also covers topics such as:
- The five questions an AIOps solution must answer
- The five types of AIOps algorithms that help answer these questions
- What’s making modern IT environments so difficult to monitor
- How the Moogsoft AIOps platform provides continuous service assurance
“I share Moogsoft’s vision of what AIOps is, what it means to apply AI to IT operations use cases,” Cappelli says.
He then explains how “to use the technologies that constitute AIOps to transform IT operations to make it better able to support modern infrastructures and the digital businesses they enable.”
Cappelli also answered questions from the audience during and after the webinar. Below we offer an edited transcript of the Q&A.
Q&A with Will Cappelli
Which APIs and custom integrations does Moogsoft AIOps offer for linking with third-party tools such as monitoring, log management, and ticketing systems?
We have more than 130 and growing. You can see all of our integrations in this section of our website.
With regards to the massive increases in IT complexity, will it become impossible for any single software to create real time alerts? Will competing AIOps companies be forced to “hang together or hang separately”?
The future will likely see a hierarchy of smart ‘filters.’ Just like the human nervous system, a massive amount of intelligent processing will take place at a central point. But what that central point receives will very rarely be raw data or even raw data with an overlay of structure.
Instead, the five algorithmic types I discussed — data selection, pattern discovery, inference, collaboration, and automation — will be applied and the outcomes passed on to another layer in the hierarchy. Eventually, however, at the top of the hierarchy, you will find a single system. From a market perspective, I expect three or four vendors to vie for the hierarchy’s top spot.
Does your platform collaborate with DataOps and SecOps?
Our platform ingests data from structured and unstructured databases and applies its algorithms to that data stream just as they would be to any other data stream. However, our algorithms are designed to understand the meaning of data about an IT environment’s behavior. Hence, the platform would probably be more usefully deployed against performance and behavioral data about the database management system itself than the data contained within that database, unless that database contained data about IT systems.
With regard to SecOps, we have customers using our platform to discover and analyze security events and event patterns. Although the algorithms are currently targeted at operational events primarily, most unearthed patterns and causality are equally applicable to security. We’ll explore this area further with our customers and partners.
How do we determine the accuracy of correlation alerts?
A correlation among alerts is a statistical hypothesis and, depending on the correlation, there are both classical and high-dimensional techniques that allow you to test that hypothesis. We give you visualization tools that help see how alerts are correlated and adjust parameters to get the desired results.
How long does the Moogsoft AIOps implementation take on average with ServiceNow on one side and Splunk on the other? How much of the Moogsoft AIOps analysis is based on rules and how much on AI and ML?
Like most complex but well architected software systems, implementation takes on average six to 12 weeks. Of course, if the environment or data sets are straightforward, it takes much less. If mistakes are made or data sets are poorly chosen, the implementation time gets extended.
Does it make sense to use a Moogsoft solution only for one part of the IT environment, such as network services?
Yes, one ‘targets’ Moogsoft at specific domains and it will apply those five algorithm types and deliver value. In a very real way, however, the value of the technology increases exponentially with the complexity of the domain/domains. Network services are usually complex environments, so Moogsoft’s value can become quite evident. However, if you were to combine network and application data sets, the value derived would be all that much greater.
What is the main difference between Moogsoft AIOps and ServiceNow ITOM? As far as I know, ServiceNow ITOM answers the 5 questions you mentioned:
- Is something happening?
- What is happening?
- Why is it happening?
- Who should care about what’s happening?
- What should be done about it?
While it may be possible to achieve each of these things using ServiceNow, it is not easy, scalable nor sustainable. ServiceNow’s platform features a form-based UI with a relational database that requires rules to stitch each “component” together to attempt a solution. It also requires a customer also have ServiceNow’s CMDB and discovery services. There is no end-to-end AI and it is entirely rules-based. This all results in a huge lift with significant effort and financial investment required, with ever more needed to support long-term.
Do you use the platform yourselves?
Yes, we use Moogsoft AIOps to monitor our environment, which is complex and data intensive. We also use our own platform to monitor all of our customers’ SaaS instances in AWS.
How do we know Moogsoft is going to work when a client’s systems are down?
The Moogsoft AIOps SaaS instances are fully redundant and we maintain seven days rolling backups at any given time.
What sets Moogsoft AIOps apart from its competitors? What’s next for Moogsoft?
Many unique capabilities differentiate Moogsoft from our competitors, including:
- Our five-dimensional approach to AIOps. Whereas most vendors only address one or two, Moogsoft AIOps covers all five dimensions.
- We filter the data first, before we do the analysis. Other vendors do not do this, so many of the patterns they discover are noisy.
- Our platform is powered by more than 50 patented AI and ML algorithms which are custom built for service assurance.
- Our Situation Room is a game changer. It’s a virtual meeting place where IT Ops teams can collaborate, share and access all the information necessary to solve a clearly-defined problem.
- We have a platform approach to AI that helps our customers transform their enterprise monitoring stack and achieve significant cost savings through integration, automation and knowledge reuse.
We continue to innovate on the AI/ML side to produce new techniques for better data analysis and correlation to help customers detect incidents faster and prevent outages. We have a number of innovations planned for this year. Stay tuned!