April 25-29, 2016

Event Details

Please note: All times listed below are in Central Time Zone

Valet: Holistic Data Center Optimization for OpenStack

Today, OpenStack schedules each virtual machine and volume almost in isolation. However, cloud applications can consist of many VMs and volumes with potentially complex requirements around relative placement. Meeting these requirements requires that holistic placement decisions be made in consideration of the entire cloud application.

In this talk, we will present our proof of concept work, Valet. It gives OpenStack the ability to meet an application's requirements while optimizing cloud resources and maximizing the use of available infrastructure. It accomplishes this through the use of Heat, Nova, and Cinder plugins, coupled with a new optimization plugin architecture.


The Telecom/NFV track is brought to you by OpenStack and OPNFV.

What can I expect to learn?
  • QoS requirements of cloud applications
  • Problems addressed by Valet.
  • Meeting application requirements/constraints while optimizing the usage of cloud resources.
  • A description of the key components of Valet and how they integrate with OpenStack Heat, Nova, and Cinder.
  • A review of the control flow when a cloud application is deployed using a Heat template.
  • Heat resource plugins: How to express QoS requirements in a Heat template. Includes VM to VM bandwidth reservations, VM properties, and affinity/diversity/exclusivity groups.
  • An example showing how a complex cloud application with 10+ VMs and volumes is deployed with and without Valet.
  • Benefits of the approach, and lessons learned in working with OpenStack.
Wednesday, April 27, 1:50pm-2:30pm (6:50pm - 7:30pm UTC)
Difficulty Level: Intermediate
Red Hat, Cloud Architect
Joe has held a number of roles at AT&T for fifteen years, most recently in Cloud Strategy, Architecture, and Innovation at AT&T Labs Research, focusing on cloud computing infrastructure and OpenStack. Previously at AT&T, Joe helped research, deploy, and evangelize a number of web, search, and content management technologies in conjunction with the then-brand-new Joe... FULL PROFILE
AT&T Labs - Research
Kaustubh Joshi aka KJ is a Lead Inventive Scientist AT&T Labs-Research in Bedminster NJ, working on cloud platforms for NFV. Before joining AT&T, KJ received Ph.D. and M.S. degrees in Computer Science from the University of Illinois at Urbana-Champaign, and a B.Eng degree in Computer Engineering from Pune University, India. KJ's expertise is in systems (and mathematical modeling of... FULL PROFILE