Circonus CEO Bob Moul recently spoke with Amy Al-Katib, Editor-in-Chief of Mission Critical Magazine, about how organizations can begin to implement more sophisticated infrastructure monitoring analytics like predictive analytics and maintenance. This is the second time in the past few weeks they spoke about how the sudden surge in online services brought on by the COVID-19 pandemic has exposed weaknesses in the state of monitoring within many organizations.

Times are changing fast and technology continues to advance at an unrelenting pace. An explosion of systems and devices, complex architectures, pressures to deploy faster, and demand for optimal performance have placed greater and greater strain on monitoring teams. For many, their current monitoring strategy and tools are just not enough.

Crash events are one of the more serious problems that can occur when operating a service. Crashing components often cause cascading failures and service outages. To reveal the magnitude of damage and help prevent future occurrences, visibility into crash events is critical. Unfortunately, debugging crashes is one of the more complicated endeavors. The state of a crashed process is often compromised and the process can’t be trusted to collect debugging information on its own.

In a recent post I talked about the strain being placed on IT Infrastructure with the current surge in demand for online services being driven by the COVID-19 pandemic. I talked about how this sudden migration to online has exposed weaknesses in, and in some cases a total lack of, adequate monitoring practices. Unfortunately, many online sites have experienced degradation of service, poor customer experiences, and even complete outages.

COVID-19 is impacting nearly every company around the world. While the pandemic is affecting companies in different ways and to different degrees, a commonality many are experiencing is that the coronavirus is forcing much of our daily commerce activity online. I wrote in a post recently that literally overnight we’ve had to find new ways of working, meeting, shopping, managing healthcare, and even staying entertained.

This week, we announced the availability of our Spring 2020 release. The highlight of the release is our Kubernetes monitoring solution, which provides health-based alerting and horizontal pod auto-scaling. Additional enhancements include cloud monitoring, GCP Marketplace availability, performance improvements, and a more comprehensive Terraform integration. Here’s some background on these latest capabilities.

Latency SLOs help us quantify the performance of an API endpoint over a period of time. A typical latency SLO reads as follows: The proportion of valid* requests served over the last 4 weeks that were slower than 100ms is less than 1%. *In this context, “valid” means that the request responded with a status code in the 200s.

A common task that users want to perform when monitoring their infrastructure is to identify their top resource consumers. Although the following techniques can be applied to numerous different resource metrics, we will specifically look at the problem of identifying which of our hosts or services are consuming the most CPU resources.

In a conversation with my sister last week, I was musing whether we would go back to shaking hands in the aftermath of the coronavirus and in general what the world was going to look like in the months and years ahead. She made an observation that I thought was spot on: “I don’t think we’re going back.” The world has fundamentally changed.

Percentiles are commonly used for measuring statistics, particularly when analyzing things like latency. Unfortunately, people frequently get tripped up when they want to take multiple percentiles and aggregate them. For example, let’s say we are monitoring a set of ten web servers and we want to collect latency statistics across all of them.