Issue 053
Hey folks, welcome to another installment of Monitoring Weekly! Did you write something about monitoring recently? Maybe got an idea rolling around in your head? Send it on over and let the community learn from you. :D
This issue is sponsored by:
Hey engineering managers! Want to improve your hiring, conduct more effective interviews, and hire great people quicker?
Your Next Hire is here to help. Check us out.
Monitoring News, Articles, and Blog posts
Protecting GDPR Personal Data with Pseudonymization
The imminent go-live of GDPR has a lot of companies frantically searching for answers. There are still a lot of questions about specifics, but this article goes into one area that is being discussed: pseudonymization. If your log data contains personal information (which is a really loose definition under GDPR), you’ll be interested in this article on pseudonymization in ELK.
How To Upgrade Prometheus 1.x to Prometheus 2.0 On Ubuntu 16.04
Are you still running Prometheus 1.x? The author of this article walks us through how to do an upgrade by keeping the old Prometheus 1.x system in place and running Prometheus 2.x alongside it, allowing you to keep using the time series data from 1.x inside 2.x.
A Guide to Instrument Sanic Application, Part 1
I really like this article for its use of actual code examples in how to instrument an application. Also, I don’t know why it never occurred to me to use Python’s asyncio to introduce artificial latency–that’s a neat idea.
Introducing Network Error Logging
Okay, so this broke my brain a little bit. Look at your Apache access log. That’s all server-side. What if the browser collected that same data, and your app could request it just it would for page timing data? Turns out, the Web Performance Working Group at the W3C is working on a standard exactly for that.
Using AWS Lambdas for data lake monitoring
Using Python-on-Lambda and Cloudwatch, the folks at Autotrader are monitoring their S3-backed data lake to ensure they have both recent data and that it’s correct. This is a pretty neat approach to the problem.
The Mon-ifesto Part 2: Alerting and Graphing
Not all alerts are equal. Some should wake someone up immediately, while others aren’t really alerts–more of information I don’t want to ignore, but don’t need immediately. And there’s several more gradations in between. How do you decide which is which, where to send them, and how? The folks at Capital One have some thoughts on that.
Graphite is pretty great, but one of the big objections to it is on the topic of scale. So I went looking for someone to really tell us how it is and found Brad Lhotsky from Craigslist. He’ll be speaking at the SF Metrics Meetup on April 4th (that’s tonight!) on scaling Graphite. For those not in San Francisco, there’s a livestream!
Full disclosure: I’m an organizer for this meetup.
See you next week!
– Mike (@mike_julian) Monitoring Weekly Editor