Summit Program

Speaker: Ted Zlatanov, CFEngine

The days of large IT organizations and maintaining systems by hand are ending. Today's infrastructure is dynamic, personalized, and malleable. IT infrastructure has become incredibly flexible and complicated as it increasingly uses cloud computing infrastructure in combination with traditional datacenters. To manage such hybrid infrastructures, organizations need configuration management solutions that enable system administrators to deploy and move workloads with ease, without disruption and with consistent behavior across the physical and virtual platforms. Ted will discuss the present and future challenges of managing IT infrastructure and will review scenarios and strategies to scale such hybrid infrastructure from a few to thousands or even millions of nodes, while ensuring reliability and compliance for all nodes throughout their entire lifecycle.

Owen DeLong will start the ball rolling with "The Cloud Can't Scale Without IPv6"; afterwards and throughout the day, additional topics will be solicited from the audience.

The speaker will deliver a presentation, which will be followed by group discussion.

Speaker: Sean O'Meara, Opscode

The world became addicted to dry-run mode when it was introduced with the make utility. Convergent policy–based CM systems are a completely different animal. The best you can hope to do in a dry-run scenario is to evaluate listed promises against the current state of the system. However, during a real-world run, operators change the underlying system, their side effects determining the behavior of operations down the line. I suspect the ability of a dry-run output to follow a linear line: A 99% configured system's dry-run output can be mostly trusted, whereas a 0% configured system's dry-run output is made up of lies and failure.

Discuss.

The speaker will deliver a presentation, which will be followed by group discussion.

Speaker: Alan Robertson, Assimilation Monitoring Project

Do you have a lot of servers to monitor? Then the Assimilation Monitoring Project may be just the thing for you. The Assimilation Project is an open source, superscale, work-in-progress monitoring project which is essentially O(1) in complexity. The ongoing monitoring of exceptions for a million servers is projected to require processing on the order of 10K packets per day. 

In addition to highly scalable monitoring of servers and services, the Assimilation project will provide Continual Stealth Discovery, which continually discovers servers, switches, services, and service dependencies without sending out packets which might set off security alarms. This will create a detailed database which will allow the monitoring system to tell you what is not being monitored and will provide significant assistance during initial configuration. Current status is that code has been written for the basic functions of the system and that it has successfully assimilated 250,000 simulated servers for monitoring in under 11 minutes.

This free-form afternoon session will be in two parts, divided by the afternoon break. It will include the following talks to spark discussion, as well as topics brought up by the audience during the morning session:

Jim Baker, Canonical, "Service Orchestration for Cloud Environments with Juju"

Vadim Spivak and Kent Skaar, VMware, Inc., "Why We Built BOSH, a Release Engineering and Service Deployment Tool-Chain"

Narayan Desai, Argonne National Laboratory, on BCFG2

Sean O'Meara, Opscode, "Finding Closures in the Open World"

See the previous session for detailed information.