Monitorama PDX 2018, in Portland, offered an intense, three-day conference program -- by and for monitoring and DevOps practitioners.
You are here
Let op! Monitorama EU Re-Debuts in Amsterdam
Back in Europe for the first time in five years, Monitorama EU 2018 brought the cream of Europe’s observability community (plus a handful of sponsors, including Opsview) to Amsterdam, September 4 and 5. The conference was held at the Compagnietheater -- a beautifully-renovated, multi-functional art and performance space overlooking the T-intersection of the broad Kloveniersburgwal -- originally a defense canal -- and the narrower Raamgracht.
Amsterdam is, of course, one of the world’s most beautiful cities, and (save for a half-day of rain on Wednesday) the late-Summer weather put everyone in a good mood. In other respects, the EU show maintained Monitorama’s tradition of prioritizing technical depth and monitoring best-practice over company affiliations and marketing agendas; espousing open source, community, transparency, and diversity; and creating a laid-back, safe place for professionals to learn from one another.
Team Opsview had a great time at Monitorama EU (as we did at Monitorama Portland, this past June). The afternoon of Day 1, we gave a brief presentation on monitoring’s scaling challenges, and some of the ways Opsview Monitor 6 has been architected to overcome them. At the booth, we shared our live demo of full-stack Kubernetes/Docker monitoring under serverless workloads and gave away many items of purple Opsview swag. On Tuesday evening, sixty or so attendees joined us for libations (including a range of what folks said were some stupendous house-brewed beers and what I can attest was the most elemental mint tea I’ve ever had) at a historic nearby inn and brewery called De Bekeerde Suster.
All this, of course, wrapped around the actual conference program, which was, as expected, the unique Monitorama mix of intensely technical and gently humane. The Day 1 and Day 2 livestreams have already been posted (these will eventually be cut into individual session videos). Meanwhile, here are some highlights (a highly subjective, and non-exhaustive list), with video links:
- Monica Sarbu, who runs ‘Ingest’ at Elastic, opened the show with an extremely moving talk about the trials of being a female engineer, calling for radical empathy and revisions in workplace culture in an attempt to stop the exit of so many talented women from tech. (video)
- Rick Rackow, of eBay Classifieds, talked about how his group switched from using Nagios, through Sensu, and finally to Prometheus, and how they were obliged to change their org’s thinking about monitoring, each time they evolved their toolkit. (video)
- Radu George and Rafal Kuc of Sematext gave a really insightful talk about how tools inflect DevOps culture, how problems like cognitive overload affect operator performance, and how social norms, politeness, diplomacy and other behaviors can assist or confound good teamwise communication and feedback. (video)
- Yan Cui, of global sports video network DAZN, offered a technically deep, highly amusing, and fact-packed tour of challenges and best-practice tips for monitoring in AWS Lambda, and by extension, in other fast-evolving serverless compute environments. A perfect bookend for Mandy Waite’s Day 2 talk, giving another perspective on similar topics. (video)
- Stephen Strowes, Sr. Researcher for RIPE NCC, the European internet registry coordination center, gave an amazingly-informative talk about how to establish and monitor internet visitor traffic from vantage points beyond your application’s entrypoints and your own domain of control. He explains how diverse monitoring solutions (including Nagios-compatible plugins) can extract data from RIPE NCC’s Atlas internet monitoring system, enabling root-cause forensics on formerly-opaque issues with DNS, security, etc. (video)
- Michael Kehoe and Nina Mushiana, SREs for LinkedIn, gave a fascinating talk on how they’re applying mature best-practice evolved by the National Transportation Safety Board (NTSB -- i.e., the people who investigate airplane crashes and similar transport system incidents) to their own work in incident management. (video)
- Marcus Barczak, of edge-cloud provider Fastly, offered his experience migrating his organization to Prometheus. Because Fastly uses a bespoke, bare-metal-only infrastructure (according to Barczak, they “embrace the 90s”), Barczak’s group needed to create a normalized API for service discovery to let Prometheus extract data from their systems. (video)
- Gregory Park and Trevor Morgan, who have collective responsibility for monitoring at Standard Chartered Bank, talked about building an infrastructure (and obtain stakeholder consensus) that let them commodify the individual, granular monitoring tools that DevOps groups wanted to use, extracting information from them into a API-accessible data lake for further analysis and integration with business-wide ops and incident management services. (video)
- Google’s Mandy Waite provided the ‘bookend’ on Yan Cui’s earlier serverless monitoring rundown. For me, the most important take-away was that future serverless computing will likely leave behind some of the characteristics of today’s “serverless functions” paradigm, becoming much more flexible and likely posing a continuous series of new challenges to observability. As she says: “Serverless can be about asking for 1000 CPUs, or a handful of CPUs and two PetaBytes of RAM …” (video)
More like this
Monitorama is finally returning to Europe! Join us again to hear talks about the newest approaches in monitoring and observability.
Having just returned to the office from a great couple of days at Cloud Expo at the ExCeL in London, it’s time for reflection and lessons learned...