The (in)famous Corey Quinn of Last Week in AWS recently graced the website of Monitoring Weekly and wrote this gem on a new/old concept: Observerless. Now with video of this talk from ServerlessConf, available at A Cloud Guru.
Have I mentioned how much I love CommitStrip? No? Well, it’s great because of how often it’s just totally on point.
The article says it all: don’t use your production database as a logging destination. Yes, there really are apps that actually do this.
I like this straightforward explanation of log levels. For those who don’t know, these are a subset of the well-defined syslog severity levels, which are awesome and more people should use them.
Another great thread from Charity Majors, echoing many of the same things I tell my clients (and have told you folks) on a regular basis. I do take a slightly contrarian view on one aspect though: there seems to be an implicit assumption that everyone is running large, distributed architectures, which I’ve found to be much less common than you’d think based on reading Hacker News and Twitter. I’d even go so far as to say that the majority of systems CAN be modeled in your head and most monitoring/observability problems are relatively straightforward.
Now, if you’re Netflix or one of the other AppAmaGooBookSoft sort of companies, observability is actually a nightmarish problem thanks to the staggering scale of everything. This episode of Software Engineering Daily features a discussion about how Netflix does observability and some of the stuff they’ve learned along the way.
Not sure what to write for your status page updates? Follow these instructions–they’re great.
Videos from Monitorama Amsterdam are now live. Wooo!
I wrote a couple weeks ago about this idea when I ran across Peter Bourgon’s article about what the future might hold. This article hits on the same ideas in more depth and presents what such a pipeline might look like.
Facebook Engineering has released a new open-source tool for distributed log storage, promising a million log events per cluster, fault-tolerance, and a high degree of availability.
Congratulations to the folks at Timescale for hitting their 1.0 milestone!
A whole bunch of examples for how you can integrate Lambda and the AWS Health Dashboard to actually do useful things.
The folks at FreshTracks built a neat tool to solve a particular problem I know you’ve experienced: unexpected explosions of cardinality bringing your TSDB to a grinding halt. This tool spots rapid increases in unique metrics and alerts you. Only for Kubernetes right now, but there’s plans for non-k8s coming.
I’m giving a teaser of my upcoming DevOpsDays KC and OSMC talk at MonitorSF in a few weeks. If you’re in SF, come on out.
Splunk is putting together a new event series that looks pretty neat, and it isn’t just a Splunk event in disguise. Best part: it’s free.
Come on out and hear my pontificate about shitty cars and bad monitoring. There’s $25 off if you use the
SEEMIKEJULIAN code at checkout.
I’m also speaking at OSMC in Nuremberg this November, which is gonna be super fun. Though, it occurs to me that “jalopy” may not translate well in German…
I don’t know how I missed this tidbit, but Grafana is launching Grafana 6 at their conference in Los Angeles in February. Also, CFP is open.
Do you have an event you want posted here? Send it on over!
See you next week!
— Mike (@mike_julian) Monitoring Weekly Editor