Why the cloud is no celebrity diet

Cloud computing is not some voguish weight-loss programme. To get the most from it requires a transformation of the whole datacentre, not just individual applications, says Lori MacVittie.

Consolidation and more efficient use of resources are behind many moves to the cloud. But for long-term success and control over costs, we need to view the move to cloud computing as a change of lifestyle and not just a fad diet.

Anyone who has tried to lose weight knows that the easy part is losing it. It's keeping it off that's really hard. That's because what is really required is a lifestyle change — it isn't just a change of diet. It's the ongoing vigilance in adhering to that diet that assures long-term success.

Trim the fat of excess resources and leave the datacentre a lean, green application-delivery machine.

Similarly, if you view cloud computing and virtualisation as merely a temporary datacentre weight-loss programme to allow you to fit into that new budget, you're ignoring the chance to transform the datacentre and its health for a lifetime.

Exploiting the cloud — private or public — requires incorporation of the model into datacentre architectures and processes. It's got to be a change in the way resources are viewed, used and ultimately managed. It is not just a quick-fix for a budget problem that exists today.

The datacentre diet

One of the goals of exploiting virtualisation and ultimately cloud computing is the efficient use of resources, particularly compute resources. In traditional datacentre architectures we are accustomed to server utilisation that is far below capacity — often less than 50 percent for any given application.

There's a lot of fat to be trimmed from the datacentre, and virtualisation combined with cloud computing enables organisations to reduce the number of physical servers while maintaining availability of applications simply by better provisioning and managing the remaining resources.

By viewing resources as an amalgamation of compute resources rather than individual compute capacities, it is possible to consolidate those resources into a single pool that can be distributed more efficiently to meet the demands of applications, thus reducing operational and capital expenditures. The datacentre can fit into that shrinking budget.

That's a good thing. That's the diet, right there. Trim the fat of excess resources and leave the datacentre a lean, green application-delivery machine.

Keeping the datacentre weight off

The trick is keeping it off, just like a diet. In the datacentre, the pounds start piling on with the scaling-out process. An application experiences a flash mob of traffic — and, by design, the infrastructure — automatically provisions the additional resources necessary to maintain availability and performance goals.

Indeed, auto-scaling is a beautiful thing, but without the proper processes in place to scale back down when traffic and usage...

...returns to normal, the datacentre gains weight. "A moment on the lips, forever on the hips", the old adage goes — and it's just as true in the datacentre.

The answer is to ensure that application usage and traffic patterns are monitored, and that infrastructure and supporting processes are in place to enforce scale-down policies, as well as the very important scale-out policies.

The application should only consume what is essential to maintain its 'weight', its operational posture and nothing more. As demand reduces, so should resource consumption.

A post in 2008 by Joe Weinman, then of AT&T now of HP, on the 10 Laws of Cloudonomics is still relevant today and summarises this point well.

Utility premium

An on-demand service provider typically charges a utility premium — a higher cost per unit time for a resource than if it were owned, financed or leased. However, although utilities cost more when they are used, they cost nothing when they are not.

This is a point often lost when we translate cloud computing to private datacentre architectures: resources cost nothing when they aren't being used in a public cloud. To achieve similar dietary success in the datacentre, we must follow the first rule of 'cloudonomics' and ensure that only resources necessary to meet performance and availability business requirements should be utilised. Supply should match demand.

The application should only consume what is essential to maintain its 'weight', its operational posture and nothing more. As demand reduces, so should the consumption of resources. It's a change in how the resources are managed, not just provisioned, and that requires more attention to strategy and policy in the datacentre to assure success.

Achieving that success requires we not only change how we provision resources but how we manage them in total. It's not just about scaling out, it's about scaling back down as well, and matching day-to-day operational resources with demand, on demand.

If we're going to be running a marathon we probably need to eat more. But the next day when we're taking a break we need to cut back, otherwise the excess resources begin to build up again until we no longer fit our budget.

Cloud computing is a technology with the power to transform a datacentre if we recognise the opportunity and develop a strategy that is long-term rather than immediate in nature. It must be a transformation of the whole datacentre, not just individual applications, if we want to achieve the efficiency and agility promised by cloud-computing architectures.

Lori MacVittie is responsible for application services education and evangelism at application delivery firm F5 Networks. Her role includes producing technical materials and participating in community-based forums and industry standards organisations. MacVittie has extensive programming experience as an application architect, as well as in network and systems development and administration.