Apples to Oranges: Why Comparing Compute Units Can Derail the Cloud Conversation

Mar 6 2012 | by Zev Schonberg

0 Flares Twitter 0 Facebook 0 LinkedIn 0 Email -- Filament.io 0 Flares ×

Two of the most popular cloud vendors today are Amazon Web Services (AWS) and Rackspace. An examination of how they bill shows that both measure the size of “compute units” differently in pricing their solutions. These compute units include variable levels of storage, memory, platform and virtual machines – levels selected and defined by the specific cloud vendors.

Amazon’s Elastic Compute Cloud (EC2)allows users to rent virtual computers to run applications – calling the system an “instance,” and there are many models for AWS costs based on reserved vs. spot instances. Outside of one time use (spot instances), for Amazon compute units, simply put, the longer the commitment an organization makes to the cloud, the lower the incremental price. Rackspace Cloud Servers are similar virtual instances that are running on Xen or Citrix hypervisors and are priced based on “server” hardware nodes.

Amazon’s dominance in the early public cloud marketplace has shaped the conversation around what a “compute unit” includes and how it is measured. The Amazon pricing approach to the market has also held market dominance and is replicated by many vendors. Pay as you go, reservation models and spot purchasing of resources are all approaches to public clouds initiated by Amazon. However, the replication of Amazon’s pricing approach hasn’t created unity in the market. For most businesses looking to move to the cloud, trying to compare Rackspace Cloud Server to Amazon EC2 when selecting a vendor is equivalent to comparing apples to oranges – there is no context for comparing one against another.

Absent a common point of comparison, many IT departments have chosen the “seemingly” simplest approach to cloud−pay as you go− with whichever cloud vendor they have an initial affinity for, without looking at their own actual organizational needs. Using this approach, the ability to make a clear comparison of Amazon vs. Rackspace compute units is the least important part of the cloud deployment decision.

Business Objectives First

In the face of dynamic resource demand, changing resource pricing models, and performance constraints, businesses need tools that map cloud resource utilization to the business objectives for their projects. This might seem like an obvious point, but many enterprises embrace the cloud before ever looking how best to utilize it and gain cost savings. To make a truly smart pricing decision, organizations need to look at cloud resource needs, performance constraints and budgetary guidelines to create a clear business model for the cloud. Without these in place, enterprises quickly face a situation of rampant overprovisioning and sticker shock at the start of any cloud deployment.

Defining Requirements of Your Cloud

Cloud comparison is also subject to other parameters as well, such as: license of software (for instance, use of a virtual server with Windows version will increase prices), support, availability, regulations, etc. Therefore, the selection of a cloud based on price should take these figures into consideration as well.

Today, it’s not really possible to do a comprehensive performance benchmark between cloud platforms. Even if an organization tries to do that, there is no “one-cloud-fits-all.” This means that even a comparison between Amazon virtual server X to Rackspace virtual server Y is done, the various platforms might not fit application performance or regulatory requirements because of factors such as location of the data center.

The specific requirements of the cloud applications you wish to deploy will help shape which cloud vendor platform will be best and is why the approach of “comparing first” using just a cost calculator doesn’t make sense. You have to follow Cloudyn’s approach of “migrating to the cloud first,” and then fine-tune your decision.

Fine Tuning Costs for Perfect Provisioning

Cloudyn’s non-intrusive SaaS solution automatically gives companies new insight into their cloud infrastructure and recommends on how to perfect their investments, reducing cloud cost while maintaining operational performance. Correct utilization of cloud for the business needs will save exponentially more than possible via a blind choice of a seemingly “good” pricing model with any cloud vendor.

Only after obtaining an analysis of the cloud environment and its relevant performance needs are businesses armed with the information they need to then look more closely at Amazon vs. Rackspace vs. other cloud vendors. With an understanding of needs (spot vs. reservation), then the comparison between vendors is more clear than by just looking at the “compute unit” pricing model.

Cloud Pricing Wars are Coming

Martin Creaner, president and CEO of TM Forum, stated in his recent “Cloud Monetization” whitepaper that, “Cloud billing should not be underestimated as a means to service differentiation. But before cloud providers can achieve that differentiation, they must first understand and be able to articulate their competitors’ capabilities and customer needs.”

Ultimately, as standardization of “compute units” becomes commonplace and the providers have a clear message for differentiation, the cloud costs will be more a factor of marketing. Cloud vendors will be judged directly on the meeting of enterprises expectations from a product and service approach. When cloud billing is consistent and expectations are clear, service providers can and will attract more customers and enterprises will have one less obstacle to cloud adoption.

As the market continues to mature, it will then be important to get a clear view of how to compare one vendor to another from a cost perspective in terms of getting an “apples to apples” comparison in price.  At this point, independent analysis from organizations like Cloudyn is something that will bring clarity to the market.

Login

SSO Login

Forgot Password?

No account yet? Register.

0 Flares Twitter 0 Facebook 0 LinkedIn 0 Email -- Filament.io 0 Flares ×