System Designer, DreamLab
Krzysztof has over 12 years of professional experience. He started his technical journey with desktop ERP applications (C++, Delphi), then moved on to webapps (price comparison service, retail) mostly in Python and Ruby. Later it was the cloud that caught his attention for about 4 years so he focused on cloud machinery and developing private cloud. He gained administration skills (and patience) while managing a Novel-based network and then FreeBSD and Linux (system of choice) clusters. Whenever possible he tries to avoid hardware-ops tasks with, however, with little success so far, and still doesn’t remember MegaCli switches. In the meantime, he tests everything available from HN. Currently, as a System Designer at DreamLab Krzysztof is responsible for PaaS orchestration and a monitoring system.
Cluster management made doable
When it comes to managing a bunch of virtual servers like one resource, the market offers countless solutions like Heat, Senlin, Cloudify, Terraform etc. Yet, when faced with a challenge to pick one, that would suit our needs we found ourselves in quite a predicament. At DreamLab we are in charge of thousands of short-lived servers and we would like to manage clusters of servers in both our own infrastructure based on OpenStack as well as in the public cloud ecosystem where we use APIs. Our search for a scalable, plugin-oriented solution that allows for regular health checks, led us to create our own application that we’re releasing for the community to use.
In our presentation, we will go through available market solutions, our own choices and finally, we will show you tools that do the trick, so you can learn from it and manage your own servers efficiently, regardless of your IaaS.