Depending on how analysts define it, the market for cloud products and services is growing anywhere from 19% (Gartner) to 27% (IDC ) per year. The growth in public cloud service usage is even more robust. At Amazon EC2, the leading cloud services provider (CSP), average daily instance launch counts grew fivefold from 2008 to 2009, and more than doubled in 2010.

A growing percentage of these customers are enterprises using the public cloud in an Infrastructure as a Service (IaaS) model. Under the IaaS model, all equipment used to support operations is outsourced including compute, storage, and networking components. With that high degree of reliance on the CSP, many organizations are seeking some form of service management to support the specific needs of their business. That need became even clearer when Amazon EC2 experienced an extended outage that took thousands of websites down in April 2011 - some for more than three days. This outage costs organizations days of lost productivity and millions of dollars in revenue.

The need to control the cloud is all the more vital when you consider that business agility and lower costs are the two primary drivers behind cloud infrastructure services migration. However, lower costs are no bargain if the service is deficient, and agility cannot exist without reliability. Adopting the IaaS model should not mean surrendering control over costs or performance or leaving one's organization vulnerable.

The problem is gaining control of your cloud environment. We have provided five easy steps to controlling your cloud environment.

Step 1: Implement your own SLAs to complement CSP SLAs
All the major CSPs offer service level agreements (SLA). The terms vary from one provider to the next, but most guarantee availability between 99.5% and 100%. Should service be interrupted, customers often receive invoice credits of varying amounts. Amazon EC2, for example, gives customers a 10% monthly invoice credit when annual uptime drops below 99.5%. This could mean as much 43 hours/year outage and still be under the declared SLA - which is the equivalent of five full trading days or up to 43 outages of 1 hour.

While SLAs do ensure that customers receive some compensation for service unavailability, prospective CSP customers should keep the following points in mind:

• CSP SLAs are "one size fits all" agreements that offer blanket credit policies for service unavailability.
• SLA credits are limited. No major CSP SLA allows for crediting customers more than 100% of service fees.
• Service credits are inadequate compensation for service outages.
• SLAs do not enable prioritization. You can't pay a premium to ensure 100% uptime for the most critical services or a discount for less critical applications.
• Implementing your own software-based SLA lets you ensure the right level of service control to suit your business needs.

Software-based SLAs give you the power to define, maintain, and enforce the fulfillment of business and compliance policies according to your own business priorities.

Step 2: Put your eggs in multiple baskets
Adopting the IaaS model means strategic and operational business decisions need no longer be influenced by the constraints of IT assets or their associated costs. In the past, organizations had no choice but to put all their IT eggs into one basket - their own in-house computing resources. The cloud frees you from that limitation but still, many companies don't take advantage of this. As the Amazon EC2 outage demonstrated, no single CSP can be relied on to distribute your services and data optimally to match specific business needs. Their business is solely focused on providing a standard service to all. If your CSP's service goes down, your business could go down with it.

SLAs should define, maintain, and enforce business policies according to the priorities you set and actively monitor service performance and make any needed adjustments before critical KPI thresholds are violated and avoid an SLA breach.

Step 3: Optimize running costs while delivering on service targets
The second most commonly cited reason for going to the cloud is to reduce costs. The cloud is a proven cost saver in terms of capital investments, operational resources, energy consumption, facility expenses, and more. However, under the IaaS payment model, running costs are still a concern. In traditional computing, we're used to the idea of paying a premium for higher levels of service. When you control your cloud, you can actually achieve higher levels of service while optimizing costs.

The secret to doing this is to define Key Performance Indicators (KPI), which can then be articulated as Business SLAs. Under the SLAs, you can assign and enforce control adjustment actions whenever a KPI threshold is breached. The result: optimized cloud services with zero additional operational resources.

Step 4: Keep a constant eye on cloud service usage
In the traditional computing paradigm, the biggest cost challenge is that initial hurdle of buying the hardware. But once you own it, you can use it without limits. As enterprises move to the cloud, there is less predictability. Usage can vary and the meter is always running. Under those circumstances, there is a legitimate concern that service costs could start to creep up. The problem is not necessarily a lack of faith in the CSP's billing statements, but a lack of visibility and granularity.

Independent cloud monitoring is a good start but not quite enough. It's better to implement tools and services that also allow you to establish limit policies on monitored resources, and to control use of cloud services to different usage groups. By setting specific limits for each group, you can ensure that you will not get a runaway CSP bill at the end of the month. SLA controls are a way to set and enforce these live policies.

Step 5: Mix and match your cloud services to satisfy business needs and get the best value
One of the most promising aspects of the cloud services model is the flexibility and agility it provides organizations to change and adapt their services to meet business goals. Many organizations do not take maximum advantage of that flexibility because they don't want the complexity of managing multiple service providers.

The answer to this problem is called an "inter-cloud SLA broker," a software tool that can provide the means to coordinate services across multiple CSPs. With an inter-cloud SLA broker, an organization can establish and manage multiple SLAs within, between, and across cloud service providers, essentially spanning cloud distribution and deployment models.

The inter-cloud SLA broker continuously monitors service health according to the KPIs and thresholds you set in the SLAs. When a KPI threshold has been crossed, it triggers an automatic notification, and/or dynamic adjustments and provisioning actions to bring the service back into line with its SLA. Such actions could include provisioning additional IaaS resources from another provider, redirecting VIP requests to the best-performing service provider endpoint, and even performing a cost analysis across cloud service marketplaces to broker the best price for an available cloud service at that time.

By enabling this high degree of automated control, the inter-cloud SLA broker gives organizations the agility to ensure that each cloud service is always applied in the right way.

In all, there are a number of ways to get smarter about cloud computing investments and it all starts by taking control and managing your cloud resources in accordance with your key business priorities. These simple steps will help to alleviate any of the complexities previously associated with the cloud.