Public, private or hybrid...

There's no question that the growth of the cloud computing market has generated a lot of discussion, creating almost as much debate and conversation as it has confusion.

But one thing is certain: cloud computing has clearly moved from buzz to business, and has changed the nature of IT delivery in the process. Just consider the numbers. According to analyst firm IDC, the cloud software market reached $22.9 billion in revenue in 2011, and will grow to $67.3 billion by 2016. Not to mention the billions that will be spent on infrastructure to support private and hybrid clouds.

The different types of clouds - private, public, and hybrid - are generally lumped under the cloud computing banner. But they are in fact different. Deciding which is best could add friction to a decision to shift applications, data, and services off-premises.

Each type of cloud has its pros and cons - making each better suited for specific organisations, scenarios, and applications. Similarly, each has an impact on application and network performance that must be considered prior to implementation. Let's take a look at each.

A public cloud is one in which the services and infrastructure are provided off-site, over the Internet. This means a company can sign up for and start using storage, processing, and other services immediately via an online portal.

As such, the public cloud offers flexibility and ease of use, making it perfect for companies that need to bring a service to market quickly, that have few regulatory hurdles, and are using data that doesn't have to be tightly integrated with other parts of the business. The public cloud also allows companies to rent compute power and storage, which is usually billed on a discrete basis.

Public cloud services are also compelling because of scale and elasticity - a service provider supporting thousands of businesses can drive lower costs than any one business alone, and can provide adaptability for changing workloads as an operating expense rather than a capital expense.

However, there are still concerns over security, privacy, and data control when it comes to the public cloud. One other major challenge is performance. Moving services to a public cloud means enterprises must accept that their applications can potentially be run from anywhere in the world - wherever the data centre of the service provider happens to be.

Most public cloud services do not specify data centre locations in their terms of service, maximising their freedom to migrate work to reduce their operating costs. In essence, the distance (and latency) in accessing applications may significantly increase for everyone in the enterprise. More surprisingly, those distances may change unpredictably.

IT organisations have not adopted public cloud computing as quickly for critical applications and data because of security requirements, integration woes, and availability concerns. In such cases, a private cloud could be a better option.

With a private cloud, businesses own and operate internal IT services that house critical applications and data inside the firewall. But they have the flexibility to shift workloads among servers as usage spikes or when they deploy new applications.

Public cloud services are also compelling because of scale and elasticity.

For businesses in sensitive and compliance-driven industries, such as pharmaceutical or financial services, a private cloud can be a highly attractive proposition. Similarly, many companies still prefer private cloud for business-critical data as it provides full control over data and applications. This helps to mitigate security and data control concerns, but is harder to scale.

Private clouds also enable IT organisations to extract greater utility from the infrastructure they already have in place. In a typical private cloud deployment, enterprises undertake the consolidation of distributed IT resources and apply virtualisation to those resources in the data centre. This enables IT to provide more cost-effective management while spinning up services faster.

However, this is a double-edged sword as private cloud deployments could put significant strain on existing resources and work processes. As IT departments consolidate resources, applications and data are typically moved further away from many end-users. Branch office employees and mobile workers are now required to access further across the wide-area network (WAN) to get the information they need. The resulting latency can often dramatically reduce performance, and make the business less productive.

At the same time, the consolidation will put more strain on the available bandwidth connecting branch offices to the data centre. With consolidated resources, many more user requests will go back to the data centre. The WAN will be responsible for carrying significantly more traffic, which could lead to bandwidth congestion, or even force the company to purchase more bandwidth.

Finally, the virtualised infrastructure and services in the private cloud may hamper IT's ability to monitor and troubleshoot applications. Virtual switches on virtualised hosts see traffic that never touches the physical network. Virtual overlay networks offer great flexibility in partitioning a network rapidly, but do so by building tunnels that traditional network monitoring tools can't see into. Fortunately, next-generation network performance management solutions are available to handle this visibility problem.

In many cases, hybrid cloud gives the best of both worlds and is becoming the standard as it provides companies with the flexibility of moving between the two models based on changing business conditions.

By spreading things out over a hybrid cloud, an organisation can keep each aspect of its business in the best-suited environment possible. The added challenge that arises when blending public and private clouds is that the integration between services becomes harder as there is a loss of data consistency. This results in added management, as well as potential differences in interface, security, processing and reporting platforms to be dealt with.

As a composite architecture, a hybrid cloud involves both of the potential networking impacts of public and private cloud implementations: applications delivered through a public service are still located potentially anywhere in the world, while private cloud applications are still consolidated into a small cluster of data centres, creating a potential network performance bottleneck.

Whether an enterprise develops a private cloud service, uses a public cloud service, or (more likely) decides on a hybrid approach, WAN optimisation is a proven way that enables organisations to attain the cost, scale, and manageability benefits of cloud offerings while gaining the performance levels and visibility needed for a productive workforce.

With the rate at which the cloud is taking off, sooner or later, more companies will need to examine how the cloud can benefit their organisations. They must consider what variation of cloud model is appropriate. Regardless of the choice, however, there has to exist a thorough understanding of the impact each type of cloud service will have on their IT infrastructure and topology, so as to ensure there is no performance degradation for users.