Interest in the OpenStack project is steadily increasing. Founded by Rackspace Hosting and NASA in 2010, OpenStack has evolved into a large community of developers collaborating on a standard open source cloud operating system. Various software distributions of OpenStack are available, and all the code is freely downloadable under the Apache 2.0 license.
Since its inception, the OpenStack Foundation has attracted more than 200 companies. The technology is known to be implemented at widely recognized organizations such as Best Buy, Bloomberg and PayPal. This article takes a closer look at the benefits OpenStack offers and explores some practical ways you can deploy it in your businesses.
Saying 'No' to Proprietary Cloud
Before looking at how OpenStack can be deployed, it's important to first understand the value proposition that it offers. Specifically, OpenStack serves as a cloud-centric software platform for companies looking to deploy their own private cloud infrastructure. Its appeal: The weaknesses of public cloud platforms.
AWS, for example, has its own application programming interface (API) and software stack, which means businesses can't easily migrate to a competing cloud provider. As you can imagine, this could be a big issue for a company developing a strategic application.
While all cloud services offer a service level agreement (SLA), it tends to be the same for all customers. In some instances, it's inadequate. In contrast, an abundance of OpenStack service providers theoretically makes it easier to find a suitable provider that offers adequate response time or predictability.
A quick look at various cloud outages makes it clear that businesses can't control when they take place - and often remain in the dark as to the severity and exact status of restoration work.
The final OpenStack advantage may be most intractable of all: Data privacy. Depending on the services offered, or the type of organization, certain data may be prohibited by law to be stored in public cloud infrastructure. While a hybrid cloud deployment where sensitive data is kept on premise could sidestep this issue, the potential for vendor lock-in and data inaccessibility remains.
Getting Started on OpenStack: Watch Your Workloads
The first thing you must do prior to OpenStack deployment is identify the workload that you intend to run using OpenStack. "Everything begins and ends with the workload," says Adrian Ionel, CEO of Mirantis, a pure-play OpenStack vendor. "Think about the use case, be very clear and have a plan for it."
John Zanni, the CMO of Parallels, says businesses should find a partner "with a proven and deep knowledge of their specific requirements" for OpenStack deployment and management. "This is a critical step that will significantly contribute to making it easier and more compelling for businesses to adopt OpenStack and reap the benefits, both in the short and long terms," he adds.
You may be tempted to modify the open source code in OpenStack for the best fit possible, but it may not be a good idea in the long run. "Don't plan a 'Franken-cloud,'" Ionel warns. Organizations that download the community version of OpenStack, "make a ton of changes" and then proceed to implement it in a way that's unique to them will "pay for it very dearly," he says.
Alan Perkins, CTO for Rackspace in the Asia-Pacific region, suggests that businesses looking to start small with OpenStack could deploy it in a laptop in a virtual machine. When it comes to a real production or internal commercial environment, though, he suggests at least two servers. "These two computers can serve as controllers with 64GB of RAM and 32GB of RAM, respectively. You add additional computers from there," he says.
[ Reference: An OpenStack Primer for IT Executives ] [ More: How OpenStack Should Prepare Itself for the Enterprise ]
Companies looking for capabilities that have yet to make it into an official distribution of OpenStack, as well as firms looking to avoid inadvertently creating a Franken-cloud, should "keep an open eye" on the OpenStack user community, Perkins says, adding, "If you feel you want to make a change to the core offering, then you can get involved in that."
Deploying OpenStack: Use Your Imagination
There are two ways to implement OpenStack. You can work through a service provider, or you can download an OpenStack distribution onto on-premises servers.
OpenStack offers tremendous flexibility, both in terms of migrating among multiple cloud providers according to changing needs and in accordance to the dictates of prices and market conditions. This flexibility, though, can make it challenging for the uninitiated to adequately visualize the practical use cases available to OpenStack.
On this front, Ionel outlines several scenarios by which you could make use of OpenStack in your organization:
- OpenStack is particularly well-suited for building any software-as-a-service applications, either as new developments or as improvements upon existing solutions.
- OpenStack can serve as a base for delivering self-service storage and service on demand to users who need IT services.
- OpenStack can be used to deliver objective storage or block storage on demand, as OpenStack Swift delivers scalable, low-cost and easy-to-manage storage.
- Finally, by switching virtual machines or services running on VMware to the OpenStack-supported KVM hypervisor, businesses can save on licensing fees.
[ Case Study: Druva Uses OpenStack as New Endpoint Data Protection Platform ][ Feature: 17 Hot OpenStack Products ]
The flip side is that businesses looking for push-button simplicity will be left disappointed. IT departments should prepare to roll up their sleeves and adopt a "DIY" mentality when rolling out their first OpenStack infrastructure. In addition, Perkins acknowledges that initial versions of OpenStack weren't easily upgradable, though he adds that "it has become a much easier task."
Kyle MacDonald, a director at enterprise data and storage networking specialist Brocade, says deploying OpenStack is more than just adopting a platform. "It's about adopting a new model for agility within the infrastructure. Starting with [new] applications lets an enterprise experience the technical and business benefits of OpenStack while also leveraging their existing application and infrastructure architecture."
According to Zanni, OpenStack enterprise adoption "will hit the same critical mass in five years that took Linux 15 years." It's perhaps for this reason that Parallels recently announced its corporate sponsorship of the OpenStack Foundation and is expected to add support for the company's Parallels Cloud Server.
Ionel, meanwhile, says he sees OpenStack as "Android for the cloud data center, adding, "OpenStack is the only open standard. There's a huge need to have something that's completely interchangeable and totally open that anybody can contribute to and adopt."
MacDonald sees OpenStack becoming "the de facto solution for enterprise private clouds." He thinks the strong community focus will let OpenStack to grow in emerging public clouds as well as in emerging solutions for carriers. "This rapid and broad growth will enable customers to deploy hybrid cloud solutions and move even faster to the cloud."
Perkins says companies should see their OpenStack deployment as an opportunity to build up something from scratch, given the cloud platform's capability to bring about savings by replacing VMware deployments. He also advocates leveraging OpenStack to establish an agile environment where continuous development and implementation is the norm.
Ultimately, it's hard to mistake the optimism surrounding OpenStack and the belief that it represents the future of the cloud. If you are interested, you can download an OpenStack distribution and get started here.
Paul Mah is a tech blogger who was formerly an IT professional. These days, he spends his time dissecting various tech news and developments at www.TechBlogger.io. You can follow him on Twitter at @paulmah or Google Plus at +PaulMah.
Read more about cloud computing in CIO's Cloud Computing Drilldown.