Autoscale and Orchestration: the Heat of OpenStack
existing monitoring and load-balancing infrastructure, and made calls to OpenStack Nova APIs as part of the scaling up and down process.
At PyCon this year, Monty Taylor, Robert Collins, Clint Byrum, Devananda van der Veen, and I caught up and chatted about what their views were of the current status of autoscaling support in OpenStack Heat. It seems that the two pieces we need the most -- LBaas and support for external monitoring systems (perhaps via webhooks) -- are nascent and not ready for prime-time yet. Regardless, Monty and his team encouraged us to dive into Heat, contribute patches, and in general, release our work for consumption by other Stackers.
Deeply encouraged by these interactions, we took this information to Rackspace management and, to quote Monty Python, there was much rejoicing. Obviously OpenStack is huge for Rackspace. Even more, there is a lot of excitement about Heat, the existing autoscaling features in OpenStack, and getting our engineers involved and contributing to these efforts.
In the course of these conversations, we discovered that Heat was getting lots of attention internally. It turns out that another internal Rackspace project had been doing something pretty cool: they were experimenting with the development of a portable syntax for application description and deployment orchestration. Their work had started to converge on some of the functionality provided by Heat, and they had a similar experience as the Autoscaling team. The timing was right to contribute what they have learned and align all of their continued efforts with adding value to Heat.
Along these lines, we are building two new teams that will focus on Heat development: one
Simply put: we're pretty damned pumped and looking forward to more good times with OpenStack :-)