(WEB HOST INDUSTRY REVIEW) — Late in June, hosting provider The Planet (www.theplanet.com) announced that it had introduced its Server Cloud product, another entry into the cloud hosting market, and a product that will work in concert with the company’s existing cloud storage offering.
According to the launch announcement, key to The Planet’s server cloud is transparency into the systems running behind the interface. The offering is built on high-end physical infrastructure, from Dell, Intel, Juniper, Cisco and Sun, along with the Canonical Ubuntu Linux OS and virtualization using Kernel-based Virtual Machine.
Unlike many of the cloud offerings on the market, The Planet’s product is priced on a monthly basis, a decision the company says reflects the nature of its customers and their needs, as well as the uses to which they’ll be putting the product.
Provisioning, however, is automated and rapid, and upgrades are seamless and similarly fast to deploy.
In an email interview with the WHIR, The Planet’s senior product manager for cloud services discusses some of the ideology that goes into designing a cloud product, as well as the decision making process behind several of the key technological choices.
WHIR: In designing a cloud computing solution, is it important to innovate in a way that differentiates your solution from other options out there? Is it important to build something sort of “industry standard” and meet people’s expectations of what cloud computing is going to be? Is it possible to do both?
Carl Meadows: The cloud is absolutely the perfect opportunity to both design solutions and to differentiate to meet the requirements of our target audience. Hosting is traditionally a business of reselling packaged hardware, and the differentiators are typically the quality of the service, data centers and network. The cloud gives hosters the opportunity to design a real product that clearly distinguishes them from the competition. It’s not just packaging.
As far as industry standards go, we strongly believe in interoperability. This expands the market opportunity by extending the ecosystem of partners and service providers who are using cloud services. Ultimately it expands and enhances the value of every cloud infrastructure provider. There’s a subtle nuance here. Interoperability is good, and the notion that clouds will be standardized offers customers and the market tremendous benefits. There’s a lot of talk about different kinds of clouds that can communicate and offer a wide swath of service and performance solutions that all work together. That’s ideal for customers who have different requirements.
The market definitions and expectations are rapidly changing and emerging. Think back just six months ago. The most important requirement for us is to meet our customers’ expectations. We’ll leverage the buzz, but we won’t follow the hype.
WHIR: In the press release announcing the server cloud offering, you make a point of describing the enterprise-level components on the back end – in particular, the Sun, Cisco, Juniper, Dell and Intel components. Can you talk about the decision to use and highlight those components versus making the infrastructure itself more abstract?
CM: We heard loud and clear from our customers that one of their primary discomforts – and their reticence to adopt cloud platforms – centered on the lack of transparency with the solutions available on the market. Our customers run their mission-critical web businesses with The Planet’s infrastructure, so they count on us for availability and performance. The concept of the unknown costs and performance didn’t jive with them.
They told us they wanted to know what they were buying when they paid us. By exposing exactly what the cloud looks like, the systems it’s running on, and exactly what they ‘re getting will help us win more business. Our goal is to demystify the cloud. The resounding feedback has been positive. We even have photos of the cloud infrastructure in our data centers so they can see the architecture where their data resides. Bottom line: customers are concerned because their past experience with cloud demonstrated both a lack of stability and inconsistent performance. By exposing the hardware platforms and explaining how resources are allocated, we’re confident they’ll be comfortable with our Server Cloud and will adopt the platform.
WHIR: You’ve also made a point of highlighting the software components – Canonical Linux and the KVM hypervisor. Can you talk about the decisions you made regarding the software components, and how they impact what people are going to see from a customer’s point of view?
CM: Our decision to use the KVM hypervisor was twofold and done so from our customers’ perspective. We wanted to offer the best infrastructure to manage, maintain and deploy for the long term. KVM uses unmodified guests, which allows us to provide customers with full root and Kernel access. This also gives the virtual machines direct access to the latest security patches and updates as soon as they are available, which enhances security. Given the massive and growing open-source support for KVM, we are confident this is the platform of the future. It also means that by basing our platform on KVM, we always have access to the most current technologies and virtualization features.
Many of our customers are active in the Linux community; in fact, more than 80 percent of our customers are Linux users with a decided interest in KVM. They’ve been very willing to speak with us about their lessons learned using the platform, so it was collaborative on many fronts.
WHIR: On the website there are 13 cloud server sizes and prices. Can you describe the process the customer would go through with regard to scaling, moving between or adding server configurations in the face of traffic spikes or rapid growth?
CM: Server sizes are a combination of virtual processors, RAM and bandwidth. Customers can change packages at any time by simply requesting an update from the Orbit customer portal. They can grow/shrink or upgrade/downgrade directly from Orbit. The process is instantaneous.
WHIR: You seem to be charging for the cloud servers on a monthly basis, whereas some people would suggest that hourly billing is pretty critical to their idea of how a cloud computing product should operate. Can you talk about that choice?
CM: Our customers are primarily web-based business running 24×7, so availability and stability are critical. They understand, as does the market, that hourly pricing comes with a premium price tag. When we polled customers, hourly billing was a low priority. They told us they wanted predictable pricing, with bandwidth that wasn’t exorbitantly priced. We’re not looking to attract customers that have huge spikes or transitional workloads. Those customers are better served by less resilient, less redundant offerings.
WHIR: Can you talk a bit about the beta process for the server cloud – how long, how many participants – and what you might have learned from the beta, and possibly any changes you made to the product during or after?
CM: We had multiple phases in our beta. In fact, the initial limited beta, which began last November, was composed of 10 very active, vocal customers. They were given early access and provided valuable feedback. The second phase was the public beta that began in February. During that phase, we deployed more than 1000 virtual servers. We learned how to manage the environment so that it scaled easily. We also had enough traffic to understand the bottlenecks, and to fully define product growth and capacity for key elements of the cloud. The data we collected resulted in changes to the platform, which ultimately yielded a better, more robust product for our customers.
