Metrics vs Logs Dilemma: Which Analytics Approach is Right for You? (part 1 of 3)

You’ve used log monitoring to get insight into the operation of your applications and infrastructure for a very long time.

Lately, you’re hearing from your friend who works in a DevOps team at a cool cloud company such as Box or Intuit, about metrics analytics.

Over there, they are running many microservices, using Kubernetes to spin up and down a huge number of containers.

Your friend just got a promotion, and he is bragging about his ability to instantly see an anomaly across his entire application stack almost immediately. Of course, you want to play with his new toy! But do you really need a new one? In the next couple of blogs, I’m hoping to help you resolve the metrics vs. logs analytics dilemma, and which one could suit your needs.

How did we get here – why the metrics vs. logs dilemma?

In data center environments, IT teams relied on log monitoring tools for a very long time. Insights derived from analyzing log data helped IT Operations to troubleshoot infrastructure issues across silos.

But the rise of cloud computing and hyper-scale companies like Google or Twitter, where Wavefront founders came from, profoundly impacted applications and services delivery.

As software moved to the heart of the digital enterprise, the availability and the performance of services at any scale became essential for the success of business.

Their application delivery and development teams could not afford to have any significant delay in knowing about their services performance degradation.

Their entire revenue lifeline depended on the optimal performance of their core services. An undetected degradation of application performance could result in an almost instant impact to their customers, SLA breaches and revenue losses.

On the other hand, an improvement in diagnosing and trending a service performance change and in particular at hyper-scale could mean millions of dollars in savings.

In this environment, a new breed of real-time monitoring at scale was born, and metrics analytics became critical.

Cultural changes in cloud computing era require sharing

In parallel with the spread of cloud computing, continuous code delivery was becoming a norm. Developers needed frequent and immediate insight into the performance of their code in production, while SREs, TechOps and cloud operations teams needed to move closer to development.

With the spread of DevOps culture of collaboration and with dissolving boundaries between engineering teams, everyone needed shared views across the entire applications stack.

The speed of sharing insights between engineers directly impacted the speed of code issues remediation, the frequency of production code updates and customer satisfaction.

In my next blog, I will walk you through several key differences between logs and metrics analytics for application performance and infrastructure monitoring.

And in my final blog, I am hoping to help you select the best approach for your monitoring needs.

Making the right choice becomes critical, in particular, if you have adopted a DevOps-driven application delivery, and need to continuously understand the impact your code is making in at any production stage and at any scale.

Watch the Recorded Webinar to Get Deeper Insight

Curious to learn more about logs and metrics analytics? Watch the recorded webinar to get deeper insight so that you can solve the metrics vs. logs dilemma!