I’m just back from Berlin and the first Monitorama EU conference. First thanks to the organisers who put on a great event and second thanks to the speakers and workshop organisers for all the great content, some of which I’ve linked to below.
Devops Weekly is sponsored by Brightbox Cloud - launch cloud servers in multiple UK datacentres in seconds…
This monitorama talk discussed the use of algorithms in monitoring and anomaly detection, and points out why some of our statistical tools are all that useful given the data we collect.
I had the good fortune to speak at Monitorama this time around, about integrating various security and penetration testing tools with a modern open source monitoring system.
Once you have lots of metrics you probably want to alert on certain states. This next presentation covered the psychology of alert design with examples from other types of alert systems.
One of the general themes of the conference was using monitoring tools for much more than measuring low level system metrics. This talk discusses all sorts of business information that is useful to measure and what tools (if any) are used for that now.
Elasticsearch got lots of love at Monitorama with a bunch of talks and workshops discussing it. This talk focused on it’s use for storing logs, looking at a number of tools for getting information in and out and some common configuration issues.
A talk about some of the limitations of common monitoring approaches, and some examples of a new clojure based toolset that aims to fix some of them. Includes a nice conceptual model of a modern monitoring setup.
A nice writeup of the two days of the conference, mentioning a few of the talks and general themes that emerged over the course of the event.
Omnibus is a tool for building monolithic system packages, where you include everything including any dependencies. It’s a great way of distributing software with complex dependencies in a simple way. This presentation explains why and how.
A good post about why, as an infrastructure engineer, you might also be well suited to running your own company or service given the high demand for infrastructure skills at the moment.
An interesting discussion of why high availability isn’t always worth the cost and additional complexity, and why many of the solutions to availability are centred around people and processes not technology.
A good introduction to getting started with Test Kitchen, an integration testing tool ideal for testing configuration management code across multiple local virtual machines.
A claim that releasing software (in this case Python packages, but it’s a general point) is normal too hard. Shows examples of the OpenStack toolchain which simplifies this, and lots of good comments with tools that can help.
Cassandra Summit Europe is an event covering training and presentations on the NoSQL database Cassandra. Held on 16th and 17th October in London, the conference will include presentations from Hailo, NetFlix and Spotify on their use of Cassandra, as well as sessions on the recent launch of Cassandra 2.0 by apache Cassandra chair Jonathan Ellis. The first 15 people to use the code DevopsWeekly will get 15% off the ticket price.
Config Management Camp is going to be held on the 3rd and 4th of February next year in Gent, Belgium. Expect discussion and participation from Chef, Puppet, Ansible, Salt, CFEngine and more. It is cunningly just after FOSDEM too so why not extend your stay for a few days?
I’m seeing lots of interest in simple roll-your-own Platform as a Service tools at the moment. AWSBox uses EC2 to make creating a node.js specific PaaS simple, based on some of the open source nodejitsu tools.
Another interesting tool in the roll-your-own PaaS space is OSv. It basically does away with the operating system layer, running a JVM directly on top of the underlying hypervisor. Currently supports Xen, KVM and EC2.
Mentioned in one of the Monitorama talks, es-utils provides a bunch of handy utility scripts for managing an elasticsearch cluster.