Adrian Cockcroft's 5 Rules For Monitoring

Posted by Baron Schwartz on Jul 7, 2014 3:51:00 AM

Monitorama was a great conference. If you haven’t yet, please check out the videos. One of my favorite talks was Adrian Cockcroft’s opening keynote, titled Please, no more Minutes, Milliseconds, Monoliths or Monitoring Tools.

What? No more monitoring tools? Heresy! Well… no. Adrian’s talk revolved around 5 goals he urged the attendees to pursue:

  1. Spend more time on analysis than data collection and display
  2. Reduce key business metric latency to less than 10s
  3. Validate your measurement system precision and accuracy
  4. Be more available and scalable than the services being monitored
  5. Optimize for distributed, ephemeral cloud native applications

The “no more monitoring tools” part of the title falls into point #1 – we already have ways to gather metrics, but a lot of efforts go little further than displaying them as strip charts. It’s a way of saying build more analysis tools, not more monitoring tools.

We have our own bias on this, too; from our point of view, the world needs more performance management tools, not more monitoring tools. But until we are a billion-dollar success, we’ll grant that this is only, like, our opinion, man.


How do we fare on the other points in Adrian’s talk? We can brag about some points:

  • No more monoliths: Our entire architecture is designed around the assumption that systems are completely dynamic, including not only the servers themselves, but also the services that run on them, the metrics they produce, and so on. For example, our installation process is the fastest and simplest I’ve seen in the industry. We can’t think of any way to make it simpler. Everything’s just magical.
  • No more minutes: We collect and analyze metrics in 1-second resolution.
  • No more milliseconds: we capture timing data in microsecond resolution.
  • No more monitoring tools: see above.

We’re not done with our solutions to some of the areas, but interestingly we’re completely in sync on the problems, the need, and the form the solution should take. It’s just a matter of engineering time and resources.

Adrian’s talk was very good, and I encourage you to watch it on Vimeo at the link above. As a poor substitute, you can also look at the slides.


Recent Posts

Posts by Topic

see all