Hello everyone !
We're glad to get back at Criteo Lab's offices for our 7th meetup.
This event will be focused on Mesosphere technologies in the production, with a special guest, Jörg Schad is a software engineer at Mesosphere in Hamburg and Florent Flament SRE, Devops @Criteo.
Please find below the agenda :
• Feedback on running Mesos in production for 1+ year
Bio: Florent Flament SRE, Devops @Criteo
Since the deployment of its first production cluster in winter 2015, Criteo has deployed hundreds of agents. We are currently running 10 Mesos clusters around the world, with 100+ production applications spread over 500+ agents.
Running a Mesos cluster has not been always quiet. Over the year, we faced several incidents leading to downtime of Criteo products. While some were due to software bugs, others resulted in human mistakes. This talk will be about some challenges that we faced, including:
* Applications traffic load balancing
* Network isolation
* Users isolation
* Progressive deployment
• Downtime is not an option - How Apache Mesos and DC/OS keeps apps running despite failures and updates
Bio: Jörg Schad is a software engineer at Mesosphere in Hamburg where he works on the Apache Mesos project. Prior to this he worked on SAP Hana and in the Information Systems Group at Saarland University. His passions are distributed (database) systems, data analytics, and distributed algorithms and his speaking experience include various Meetups, international conferences, and lecture halls.
Many talks and tutorials show how to deploy an application onto a cluster system, but don’t talk about the “morning after”, which is about keeping and maintaining that application in production.
There are a number of challenges involved, including:
• System failures
• System upgrades
• Application failures
• Zero-downtime application updates
• Varying workloads.
This talk will detail how DC/OS, the Datacenter Operating System which is built around Apache Mesos, ensures that your application stays up and running despite those challenges.
We will discuss the following topics: How DC/OS keeps itself running despite unreliable infrastructureHow we can update DC/OSHow DC/OS keeps applications up and runningHow we can update applications with new versions and configurationsDuring this discussion, we’ll look at the technical architecture of DC/OS and Apache Mesos, and talk about best practices for development and operations.
Looking forward to see you there,
The CNCP organization team.