Datadog is an infrastructure and application monitoring vendor. What that means is that when something goes wrong in your increasingly complex IT environment, Datadog tells you its gone wrong and allows you to take remedial action. So far, so good but there are two problems with that proposition. Firstly, I really want my monitoring vendor to learn and tell me when a problem is likely to arise rather than reactive tell me when one exists. Secondly, I don’t actually want to be notified that it’s found a problem, I want it to do whatever it takes to put that problem right. Automatically.
While the second problem is a difficult problem to solve, the rise of machine learning and artificial intelligence makes the first one less so. To this end, Datadog is today announcing Forecasts, a new platform feature that promises to predict when performance and stability issues will arise within cloud applications. And the promise of Forecasts is a lofty one. Datadog is suggesting Forecasts will alert a team “days, weeks, or months before an issue occurs.”
The increasing complexity that occurs in modern IT environments has made a product like Forecasts all the more important – in highly distributed and dynamic cloud environments, planning for future performance issues and avoiding downtime can be difficult and labor-intensive. By applying machine-learning algorithms to massive amounts of data, Datadog is promising that it is able to generate predictive analytics on everything from application performance to custom business metrics.
This product is the personal baby of Brad Menezes, Datadog’s Product Manager for artificial intelligence and machine learning. As such it should come as no surprise that he is effusive both about the opportunity and value driven by Forecast, and how well his team has executed the product:
Today, DevOps teams often receive critical alerts after their customers have been negatively impacted. Our forecasting algorithms have been fine-tuned based on trillions of data points across hundreds of thousands of servers daily. We can predict where a metric will be in the future, taking into account historical patterns, and notify users with plenty of time to prevent any negative impact. This directly translates into dollars saved and better user experiences – something every organization needs.
MyPOV
As I explained at the start of this post – there are two things I want to see from monitoring vendors: predictive analytics and automatic action-taking. This move from Datadog gets them halfway and, from that perspective, is a positive step. I’m looking forward to them meeting my other requirement for a true “auto-pilot” approach towards infrastructure and application management.
