[This blog is available for download. See the pdf article here].
Today, the pace of IT is accelerating at an incredible rate. As businesses digitalize, they are demanding a vast range of new applications and services from IT, along with the agility needed to deliver and evolve quickly. The cloud and DevOps are enabling this fundamental shift, but they are also dramatically increasing the speed and volume of change – and driving an explosion in operational data. For example, modern microservice-based cloud applications generate more than 10 times the monitoring data of their predecessors, creating an unprecedented flood of events for IT operations.
Against this backdrop, some AIOps vendors advocate taking a fresh start – relying entirely on artificial intelligence and throwing away existing tools such as the CMDB. They argue that current ITIL tools and processes are built for yesterday’s static IT environments, and that digitization and the cloud demand a new approach. Rather than combining the power of AI and the CMDB, they try to convince their audience that AI can solve everything – which it can’t. Often, this position is born out of necessity, since these vendors have nothing else to offer.
Debunking the Myth of the Static CMDB
First of all, let’s address the claim that a CMDB can’t keep up with today’s IT environments. While it is true that a manually maintained CMDB struggles to keep pace, we’ve moved beyond that. Today’s automated discovery tools do more than simply discover assets. They discover applications, relationships, and even complete end-to-end business services. All of this information flows into the CMDB, and it’s increasingly real-time as cloud vendors adopt an event-driven configuration approach. For example, the Amazon AWS Config API directly represents cloud resources as CIs, and it sends notifications when a CI is created, deleted, or modified.
DevOps Drives CMDB Maturity
And, let’s be clear. While the cloud is a highly elastic environment, with developers spinning up ephemeral workloads rather than relying on static application architectures, it’s not the Wild West. In fact, these “infrastructure-as-code” approaches actually increase discipline. It’s not about blindly creating virtual machines and running software. Instead, modern DevOps methodologies use tools such as Puppet and Chef to validate and manage configuration, extending configuration visibility above the cloud and into the application layer. This even applies to monitoring systems, which are included in the automated provisioning and configuration process.
The CMDB Is Here to Stay
So, why on earth would you abandon the CMDB when it’s the future – a key point of alignment for the industry? Replacing this with a magical AI box makes no sense. And, when you consider how the CMDB plays a critical role beyond IT operations in areas such as GRC, application portfolio management, and – increasingly – security operations, it’s not going away anytime soon.
Do You Need a Mature CMDB to Take Advantage of AIOps?
No, you don’t. There are enormous advantages to having a mature CMDB – we’ll get to that in a minute – but you can still get significant value without CMDB data. For example, Evanios uses a combination of methods to correlate events and metric data, including event content, occurrence time, source, device type, preconfigured logic, and artificial intelligence. None of these require a CMDB. And, taken together, these methods significantly reduce event noise and make it easier to diagnose and remediate service and infrastructure issues.
However, here are just some of the benefits you get with a CMDB – and, as your CMDB matures, these benefits continue to increase.
- Intelligent Noise Reduction
When a failure happens, events cascade across your business services and IT infrastructure. CMDB relationships let you see how events propagate, which is critical for filtering and correlating these events. For example, parent-child relationships can be used to create event hierarchies, so that symptoms related to the same underlying issue are grouped together. And, this isn’t limited to directly related CIs – hierarchies can span multiple relationships, and even take into account other CI attributes such as location. Compared to correlating events without CMDB data – for instance, by using source or timeframe – Intelligent event management systems provide a much deeper level of noise reduction.
- Accurately Prioritized Work
Even with intelligent noise reduction, you’re still left with hundreds or thousands of events to work on. Which ones do you focus on first? Which ones are most critical to the business? To prioritize events, you need CMDB data. For example, CMDB service maps can show you which business services are affected. Key CI traits such as relationship counts can be used to predict the likely impact of an event. Other ITSM data – for instance, similar previous incidents – can be used to determine likely severity. By combining AI and CMDB data, intelligent event management systems automatically take all of these factors into account, giving you a numerically prioritized list of events.
- Automated Root Cause Analysis
When there is a service outage or degradation, IT operations teams typically spend a large amount time trying to identify the root cause. This results in extended outages and increased costs, creating a negative business impact that can run into the millions of dollars. Intelligent event management systems automate this detective work, giving you a ranked set of probable causes. To do this effectively, they need CMDB and other ITSM data. For example, CMDB relationships are used to separate causes from symptoms, and incident management data shows the historical probability of each potential root cause. And, remember that changes are a leading cause of service issues – and these are tracked in your ITSM platform as well.
- Prediction and Prevention of Service Issues
Intelligent event management systems do more than help you resolve existing service issues – they can also predict service outages and degradations before they happen, so you can take action and prevent them. They do this by learning subtle event patterns that typically lead to service issues. Again, CMDB data significantly enhances this prediction capability, allowing the system to detect patterns that span multiple related CIs. And, just like with other events, CMDB and other ITSM data makes it much easier to automatically prioritize and diagnose these predicted issues.
- Cross-Stack Visibility and Control
Ultimately, CMDB data gives you visibility and control across all of your technology stacks. Yes, it’s possible to correlate application, network, system, storage and other monitoring data without a CMDB, but that correlation is limited. By leveraging CMDB relationships, intelligent event management systems provide deeper and more accurate correlation and root cause analysis, dramatically reducing war-room finger-pointing and accelerating MTTR. And, by wrapping CMDB and other ITSM data around events and metrics, they provide business context – creating situational awareness and ensuring that you work on what matters.
The Bottom Line
If you don’t have a mature CMDB – or don’t have a CMDB at all – AIOps can still help you to reduce event noise, and diagnose service and remediate service issues more quickly. We see this at Evanios every day. However, CMDB data delivers dramatic additional benefits, allowing you to scale to today’s ever-increasing operational demands. When a vendor tells you to give up ITSM data and adopt a greenfield AI approach, you’re getting poor advice. With Evanios, you can grow your event management capabilities as you continue to mature your CMDB – there’s no need to throw everything away and start again.
Evanios uses CMDB, Incident and Change data to enrich and process events