Hybrid marches on to standardisation

Summer may be starting to slip away but it is still the season for surveys. One of the latest, from hosting company, Rackspace, highlights the fact that there is both growing maturity in the cloud services and delivery marketplace and a standardised approach starting to appear.

The study, conducted by research specialist Vanson Bourne, investigated the use of different types of clouds – public, private and hybrid – along with dedicated servers by UK and US enterprises.

It found that 60 percent of respondents have moved or are considering moving certain applications or workloads either partially (41 percent) or completely (19 percent) off the public cloud because of its limitations or the potential benefits of other platforms, such as the hybrid cloud.

The research also showed that 60 percent of IT decision-makers see hybrid cloud as the culmination of their cloud journey, rather than a stepping stone to using the public cloud alone for all their cloud needs.

`Public cloud’ can still be one of those terms with multiple interpretations, but according to Nigel Beighton, VP for technology with the company, the Rackspace definition a pay-as-you-go, multi-tenanted environment that users self-manage.

It is easy to see why a public cloud environment would not suit all applications and services a business might want or need, so much so that it now seems possible to predict that hybrid cloud will soon enough become the de facto standard cloud model.

“That’s exactly right. That is exactly where we are going,” Beighton said. “The interesting thing about this survey is that it shows the market is maturing, and when markets mature people start optimising what they do.”

And as users start optimising within a hybrid model common patterns will start to appear, which is allow increasingly standardised operational models to be developed and utilised. This process is already happening and Rackspace already runs an advisory service with a set of guidelines available to its customers.

“There are three standout situations that form the basis of those guidelines, He said. “The fundamental one is that some technologies run better in one environment than in another. For example: if I am running a Mongo database I want to have all the memory in the machine, and not on any shared environment. But the extracted data that I want to analyse I can export to a data warehouse running in the cloud. This is because analytics is very CPU-intensive, so I want as many CPUs as I can have. That way I get the best of both worlds.

“The second situation is cost. If the work is not going to peak or fluctuate too much, and does not require anything to be spun up or down, then it is more sensible to run it on the bare metal of a dedicated environment. But if work has peaks in it which require scaling up and down, then it makes more sense to run it in the cloud on a pay-as-you-go basis.”

So the job of the advisory service is to model the customer workloads and put all the constant capacity tasks onto a dedicated, managed service, and the variable tasks onto public cloud services.

“The third one is the need to protect a customer’s intellectual property and other data sensitivity issues such as PCI transaction data. Some will think there is no point in taking any risks with that type of data in a public cloud. In the end, people have a mixture of both sensitive and non-sensitive data which should be split between private and public, and that is exactly what they should do.”

Follow these trends out and it does end up with the ability of CSPs like Rackspace to start configuring platform templates that can be applied to types of business requirement.

Rackspace is already doing this as a beta tester for the Heat open source project, which is a new part of OpenStack that is designed to orchestrate multiple composite cloud applications. It is due to formally appear in the next release of OpenStack, codenamed Grizzly, which is scheduled for release towards the end of this year.

Heat offers some interesting potential for cloud service developers. Not only does it allow them to develop service templates to run on a growing number of OpenStack hosting services, it also allows applications and services that have been developed to run on Amazon’s AWS CloudFormation template format to be moved to an OpenStack environment by integrating with the OpenStack APIs.

This means that businesses can utilise Amazon’s AWS as it was originally designed to be – a development platform – and then port the service to other service providers that will be, for reasons such as data sovereignty issues, more appropriate as a host for a production environment.