I have noticed an evolution in software APIs for not only connecting Business Intelligence (BI) tools but also with a more widespread use of dashboard capabilities and the ability for customers to build their own User Interfaces (UI). This has been coupled with the popularity of time series databases for trend and behavior analysis. I find that in the area of AI and Machine Learning, a time series of events can be analyzed from a historical point of view, a present tense (state) and a predictive future state. One of the growing popular packages is InfluxDB with a Grafana dashboard.
In the days of the past, building a UI for software involved work by full stack developers in JavaScript, database abstraction via APIs, and potentially packages such as AngularJS. Building a single pane of glass across a number of disparate data sources for a combined dashboard could be even more daunting.
Today that paradigm has changed. A person can go out and use a package such as Grafana, an open source analytics and visualization suite, to build a quick dashboard in less than a day. As the package does not require any software development skills, the package configures easily by firing up the Grafana package and adding data sources and plugins. Grafana ships with a data source for InfluxDB, Elasticsearch and most of the popular SQL database packages. There is a set of default plugins for visualizing the data (e.g. tables, charts, graphs…). There is also a rich community of contributors to the plugins and data sources known as Grafana Labs. There are also even more to be found in Github among a community of contributors.
One area where I have seen UI adoption of this kind is in the area of Network Management. I see companies building what most people call a “single pane of glass” for their network operations staff. Instead of having four or five different network management tools to log into, each with their own proprietary User Interfaces, people are combining and aggregating the important data from each tool into a single Grafana dashboard (or set of a few context-oriented dashboards). Network management vendors are beginning to publish data sources and APIs to enable customers to easily customize views into their tools.
One example is for a company called Statseeker that delivers a network management solution to some of the largest enterprise networks in the world (Statseeker scalability is amazing). They publish a data source on Github and make it available for any customer to download and use in building their own Grafana dashboards.
I have seen examples where these large enterprise customers are building Grafana dashboards for multiple Statseeker virtual appliances (globally distributed) along with other vendor network management tools to aggregate into a “single pane of glass” dashboard view. Statseeker also provides examples such as “Table as a Time Series” visualization to again support the notion of displaying time series event data for critical interface state changes and resource management. Other vendors are also starting to support this trend. The Grafana testimonial site has some great case studies of large enterprises following this new trend.
Grafana is not the only player in town to support this new trend in analytics and visualization (including support for time series data), however they are one that caught my eye. I am using them to help visualize time series event data for a light industry AI IoT cybersecurity (Machine Learning) project I am working on. This includes leveraging noSQL time series data stores like InfluxDB. Any tool that enables rapid development and prototyping for data analysis and visualization is valuable and allows me to focus more on the business or cyber-problem I am trying to solve with AI.