Cloud Bursting With OpenShift

It’s 8PM, do you know where your true geeks are? Well, the room was packed for the Cloud Bursting with OpenShift birds of a feather session so the Red Hat Summit 2013 in Boston might not be a bad guess.


Grant Shipley lead the session by skipping the “Intro to OpenShift” because 18 out of 20 people in the room are already diving in. Awesome. Grant runs his blog on OpenShift making it one of the few that can elastically scale in response to load.


OpenShift is an ideal platform for deploying and hosting web applications because of its architecture. Supporting high deployment density is second nature and the ability to scale out (or in) automatically is just gravy. When Grant’s blog isn’t taking any traffic, his OpenShift platform consumption is effectively zero. But, under load, he could have hundreds of instances of the blog app pumped up on the OpenShift hosting servers, without having to touch anything or even worry about it.

The public OpenShift actually runs on Amazon EC2, which was an intentional decision on the part of Red Hat at design and development time. They wanted to create a platform which was inherently cloud scalable and deployable on local, hosted, and third party architecture. Since OpenShift first went into public beta 2 years ago (at Red Hat Summit 2011) there have been over a million applications created on the platform. Now, Openshift is available for paid, enterprise services as well as self-hosted deployment.

For a demo, Grant showed the HA Proxy status page while performing a web-base load test on his OpenShift application. You could see as the utilization reached a certain threshold that new instances were automatically provisioned and taking traffic. A minute after the requests stopped coming in, or started coming in more slowly, extra instances were de-provisioned. This use case is great for applications which peak and fall in terms of demand over time. Even when the time scales are measured in terms of minutes, the OpenShift architecture helps you ensure that your servers are giving you their best, but only when necessary.


Hosting production applications on OpenShift is as easy as developing on OpenShift, but there is some cost involved. However, the pricing model starts off very low because the architecture is capable of high vertical density. To run an application using 3 Gears 24×7 (or 2,280 hours a month) is free. What’s a Gear? A SQL database is a Gear, an application is a Gear, and any intermediary kind of platform is also a Gear. Considering common 3-tier architecture, you could say “your first app is free, but the next one might cost you”. What a great excuse to get started.


As was publicly announced at the Red Hat Summit, Accenture is standardizing web deployment on OpenShift. However, Accenture isn’t the first. Already Web PaaS providers are popping up, such as GetUp Cloud in Brazil. I’m sure I’ll be entering the fray quite soon myself with Shadow-Soft.


Not all containers are created equal. OpenShift does advertise a container architecture, but it is important to note that “container”, in this case, is not a proper noun. It does not refer to “Linux Containers”. Instead, it is a classical (and clever) configuration of CPU Groups, SE Linux policies, and permissions which come together to provide containment to individual user and application tenants on the OpenShift hosting platform. When you install OpenShift yourself, you can use these techniques to say “user ‘x’ has a minimum of 30% memory and CPU before scaling starts” and other such logical scaling boundaries.


OpenShift is very active right now, primarily on IRC via the #OPENSHIFT_DEV channel on FreeNode. Generally, members of the development team are always online somewhere in the world and a conversation is flowing between users and developers. Try getting that with commercial proprietary software. :–)

Closing Thoughts

OpenShift is the most exciting thing Red Hat has going right now, in my opinion. I am eager to work with it on premise. The advantages it gives in deployment scalability for web applications (or, can you say, Web APIs?) is a blast from the past of hand tuned high performance web hosting services with all the benefits of modern application and user separatation and on-demand (on HTTP request) elastic scaling.

Also, the My app runs on OpenShift track jacket I picked up for deploying on OpenShift live while at the Summit is sweeeeeeeeeeeeeeeet.

Related Posts

Filled under: Uncategorized