One thing that really annoys me is marketing people using buzzwords that they don’t really understand. Another is people who drive Toyota Priuses and don’t signal. But I digress…
The two most abused words right now in IT are “analytics” and “intelligence,” which are normally assigned to products (and vendors) that exhibit some kind of dashboard. You know the ones with big pie charts, colorful bar charts, and line charts with interesting spikes. To the average muppet in tech marketing, these widgets represent analytics, and the interesting spikes represent some kind of intelligence that is apparently being delivered to the user.
The Real World
This all looks and sounds great, but in the real world these dashboards deliver minimal value in the operational trenches. Why? Because most users still have to interpret, analyze and understand the data being shown – before they can actually do something useful with it. Dashboards also imply that users are watching them 24/7. This does not happen as most enterprises have between 50-100 single pains of glass.
Let’s take a look at the definitions of “Analytics” and “Intelligence” according to the proper English dictionary. Note: Wikipedia is updated by marketing people, so I didn’t use that.
Analytics – noun
- (used with a singular verb) Logic. The science of logical analysis.
- (used with a singular verb) The analysis of data, typically large sets of business data, by the use of mathematics, statistics, and computer software.
- (used with a plural verb) The patterns and other meaningful information gathered from the analysis of data.
The key phrase above is “the analysis of data by the use of mathematics, statistics, and computer software.” Computer software that simply collects data, indexes it, stores it and makes it available for retrieval via search, charts and dashboards IS NOT analytics. For example, an SQL statement is NOT analytics in the same way that a Splunk or New Relic query is NOT analytics; all of these represent questions that are asked by users versus the software asking the question implicitly. Users must know what questions to ask – this is bad because it assumes that all users know what to look for in their mad world of application and infrastructure complexity. I can assure you this is NOT the case. If everyone knew what questions to ask, or what to look for, IT operations would solve problems faster than Donald Trump.
Analytics is about software analyzing data, asking the right questions, and making users aware of the meaningful patterns or insights that exists within, so users can make faster and smarter decisions. Software analytics shouldn’t depend on the user to ask the right questions, or require him/her to define what is normal or abnormal, it should be able to learn based on the data sets and environments it analyzes.
For example, collecting thousands of events or alerts every minute from your applications and infrastructure, and presenting that data in a dashboard isn’t analytics. The dashboard may look sexy and have beautiful widgets…but…this isn’t analytics. Users applying filters on this data also isn’t analytics because it’s the user that is performing the analysis and work. However, software performing de-duplication and correlation of those events automatically so users can make sense of why those events happened in the first place – this is where you start to automate the analysis of data, and start delivering real analytics, insight and time savings to users.
Intelligence – noun
- Capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc.
- Manifestation of a high mental capacity: He writes with intelligence and wit.
- The faculty of understanding.
- Knowledge of an event, circumstance, etc., received or imparted; news; information.
- The gathering or distribution of information, especially secret information.
Simply put, can the software/product learn and understand about the environment it exists in, so it can provide unique insight to the users it serves? ‘Learn’ being the key word in that sentence. Deriving intelligence from applications or infrastructure shouldn’t require an end user to define what an application or infrastructure is, where it is, or what KPIs are important, or what normal or abnormal is. This is not intelligence because a user is telling the software exactly what data to show.
If a single pain of glass requires a user to define everything in it, then this too cannot be classed as intelligence. This user definition is better known as configuration or professional services in the world of IT, and its expensive. Custom dashboards aren’t the Silver Bullet
Customizable dashboards sound great and demo great. Unfortunately, they don’t work in the real-world because applications, services and infrastructure are so dynamic. Unless these entities are auto-discovered, modeled and visualized using analytics and machine-learning, all associated dashboards will be out-of-date before the first metric is ever rendered to the end user. Hello microservices!
I had a conversation with a chap in IT Operations last week at a conference who built several “Service Intelligence” dashboards using a log monitoring tool. I asked him, “what happens when development creates a new service?” He replied, “Well I need to define that new service and all its KPIs in one of my existing dashboards.” Not exactly scalable in your typical enterprise, I immediately thought.
Scale and Change Tolerance
Single Panes of Glass must be scale and change tolerant. They can’t rely on end users to define and maintain dashboards every time an application infrastructure grows or changes. The more IT software relies on humans to define the madness and complexity around them, the less intelligence and value they actually deliver. Analytics and Intelligence are about discovery, analysis, learning and deriving insight from data, all without human intervention.
The only way you can cut time and cost from IT is to automate what users spend the most time on.
About the author Steve Burton