Work at the OpenStack Summit in Portland this week is helping bridge the gap between what’s been promised and what’s becoming possible. Nowhere is the gap between these two wider than with autoscaling. Everyone expects the cloud to be able to automatically scale right out of the gate, but the truth is that it requires a lot of work and sometimes only happens with the help of a third-party service provider.
Making autoscaling work, and fulfilling cloud expectations starts with consensus. The challenge for this Summit was for the various contributing partners to come to an agreement about how to integrate autoscaling into OpenStack.
Should autoscaling stay part of the Heat project, which encompasses all of orchestration, or spin off as a separate project? Is scaling fundamentally different from configuration or simply a subset of what’s encompassed in workflow management? This is a major issue and gets to the heart of how people use cloud computing both today and in the future.
With the discussions today, the community has come together and recognized that Heat is more than a service for providing pre-configured deployments. Instead of just a project, Heat is emerging as a toolkit, a container for all orchestration systems across OpenStack. Autoscaling will exist in Heat, but as a separate codebase with its own API, accessible with or without implementing Heat.
This highlights the benefits of a collaborative and open technology environment. How remarkable is it that software engineers from HP, Red Hat, IBM, Rackspace and others can come together and discuss both the needs of customers and the technical implementation of the solutions? The Summit brought together multiple perspectives in a collegial environment to make real progress.
This architecture makes it easy for users to start using autoscaling immediately, without having to adopt a full template-based orchestration system. For some this simplifies their infrastructure. For others, it eliminates the need for implementing a second orchestration engine on top of one they may already have. But it also keeps the key autoscaling functionality as part of the overall orchestration engine for those who want an end-to-end OpenStack implementation.
Autoscaling is a critical part of the promise of cloud computing. The work at this week’s OpenStack Summit is going to bring this functionality into the hands of open cloud users everywhere.
Felix Sargent contributed to this article.