Run Like Rackspace: Continuous Updates

Filed in Cloud Industry Insights by Adam Dawson | April 2, 2013 10:00 am

In the cloud era, the legacy model of software updates is broken. Vendors selling five years of support services for a point-in-time release of their cloud platform are doing it wrong. Companies offering to customize a forked version of a cloud software stack for an individual customer are doing it wrong, too. In order to achieve true business acceleration[1], Run Like Rackspace through constantly updated, rigorously tested cloud software backed by world-class operations and support.

In the legacy IT model, system administrators scheduled regular maintenance cycles that looked something like this:

  1. Direct traffic away from a physical server
  2. Update the server’s operating system with the latest patches and reboot
  3. Manually spot check the server after rebooting
  4. Update the application software to the latest version and reboot
  5. Spot check the application for errors
  6. Re-route traffic to the server

Some software vendors are dragging this legacy model kicking and screaming into the cloud era. Long release cycles, manual upgrades and scheduled downtime are built into the plan for these products.

Rackspace, on the other hand, operates one of the largest, highly-available, continuously updated clouds in the world. Since August 2012, we have pushed over 1,500 updates to our cloud software with no customer disruption.

Stay close to the trunk

By using trunked versions of community OpenStack code, we can ensure that no customer or application is individually affected by an unexpected update. Clouds that use a customized, or “forked,” version of the underlying code must be individually tested and validated against each update, and the inevitable result is that the software simply becomes more and more out of date over time. This means that the end user loses the benefits of new features and fixes as they are released. By keeping our cloud on trunked code, Rackspace can bring all the latest features of OpenStack directly to customers and can help customers avoid vendor lock-in because of customized, out-of-date, “crazy cloud[2].”

7,000 automated unit tests

Operating a cloud at scale requires automated testing. Before a production release, Rackspace engineers perform over 7,000 tests on every code change to ensure nothing breaks when the code is deployed. If your cloud is intended to serve multiple applications and grow with your business over time, you need the benefit of automated unit testing to keep your cloud up to date. Rackspace’s expertise in operating a cloud at scale can help your business upgrade your cloud with no application disruption and without worry. Can your internal IT administrators or your cloud software vendor provide this kind of rigorous testing prior to pushing a new release to your cloud?

Run Like Rackspace

When you adopt cloud, make sure you are adopting a highly automated, standardized cloud infrastructure that enables you to Run Like Rackspace[3]. Make sure you aren’t stepping into the legacy IT model with six- to 18-month release cycles with scheduled disruptive maintenance windows and manual testing. You can have a cloud that benefits from the expertise of our teams running cloud at scale. You can have a cloud with new features and fixes that does not disrupt your application. Your cloud can Run Like Rackspace.

Endnotes:
  1. true business acceleration: http://www.rackspace.com/blog/run-like-rackspace-accelerate-your-business/
  2. crazy cloud: http://www.rackspace.com/blog/run-like-rackspace-an-open-cloud-in-your-data-center/
  3. Run Like Rackspace: http://runlikerackspace.com/

Source URL: http://www.rackspace.com/blog/run-like-rackspace-continuous-updates/