Rackspace’s OpenStack shows the way for public cloud vendors
Ed Albanese leads business development for Cloudera. He is responsible for identifying new markets, revenue opportunities and strategic alliances for the company.
Rackspace’s OpenStack announcement is an important step in expanding the utility of public clouds. It will push other cloud vendors to stop “forking around” and enable the standards customers are actually using. Rackspace’s move is a win for consumers and for Rackspace – it creates the opportunity to run industry standard software in truly optimized ways.
Open source software business models and public cloud business models, thus far, have been somewhat incompatible. In some cases, public cloud vendors compete with open source software vendors by offering the software vendors product as a service and delivering implied or actual support as a bundled component. There has been a fair bit of discussion on this topic. Of course, that’s a business problem – not a user problem.
Increasingly, however, the precedent has been for public cloud vendors to take open source software and “fork” it to optimize it for their specific infrastructure. A “fork” of open source software is when code is changed, and those changes are kept private, creating multiple incompatible versions of the original package. In the process, they deviate from the open standard that application vendors, system integrators and in-house developers write against. Critics would argue that public cloud vendors do it to create “lock-in”. Public cloud vendors argue that it allows them to optimize for their infrastructure. Either way, this is a user problem.
A fork from the standard forces everyone – ISVs, system integrators, in-house developers – to account for these differences in their respective applications and either modify them or develop against only part of the standard – the absolute lowest common denominator. This is the wrong approach. It stifles the ability of customers to get applications optimized by the vendors that know it best. It locks them into deployments and commercial relationships that may not be in their best interests as requirements and the market evolve. I believe this has been an inhibitor to real production deployments and wider adoption of public clouds.
The benefits of open source for cloud customers are no secret. Some forward-looking vendors – Eucalyptus, for example – recognized those benefits early, and announced fully open source implementations of popular cloud infrastructure and APIs.
Today, Rackspace Hosting takes it one step further. The company has announced the OpenStack project, which combines the Cloud Files and Cloud Servers software used within Rackspace’s internal hosting infrastructure (about 300,000 servers) with technology from NASA’s Nebula Cloud platform. OpenStack is now freely available and distributed under the Apache 2.0 open source license.
OpenStack is a positive move for both prospective and active users of public clouds, for other cloud providers and for software developers everywhere. No doubt, the ongoing success of the Eucalyptus platform, which is aimed more directly at private and hybrid clouds, caught the eye of Rackspace. Eucalyptus has it right; if you want to deliver a cloud infrastructure, you need to work well with a wide variety of applications. The key to doing this not just well – but the best ways possible – is to engage the experts; the software vendors and developers of these applications and services. There are multiple ways of doing this – but open source is the most effective in my opinion.
In preparing for the announcement and roll-out of OpenStack, Rackspace has invited Cloudera to participate by making Cloudera’s Distribution for Hadoop work well on the platform. I’m excited about this collaboration for three reasons:
- No one knows how to create and commercially enable Hadoop better than Cloudera. Cloudera will absolutely make sure our customers who elect to run on Rackspace’s platform have a great experience and that our tools do what they should. This should be much simpler with access to the complete source code that powers the infrastructure.
- Hadoop is a critical component to many successful PaaS environments. If you are going to build a new web application, you are surely going to want to understand how it is performing and analyze the data you are collecting and generating in new ways. There is no shortage of public and private cloud environments that leverage Hadoop as a key analytics component. Those that run on OpenStack will no doubt benefit from the optimizations Cloudera and others will contribute to ensure the standard works well
- A new vector in public cloud competition has been introduced. By delivering a platform where the widest variety of applications has been / can be optimized for deployment by application vendors themselves, closed source infrastructure vendors will be at a disadvantage as application vendors optimize where they can and run slower and less reliably where they can’t.
Disclosure: Rackspace is a customer of Cloudera’s. They use our software as part of their infrastructure to monitor and track performance and troubleshoot issues.