SCIENCE

Cloud computing is either a revolutionary trend sweeping through the information technology world, or an evolutionary development growing out of longtime trends. Or maybe it's both. In terms of buzzwords, cloud computing carries echoes of many previous technology and marketing phenomena, including network computing, utility computing, distributed computing, and virtualization.

With cloud computing, software and more basic computing services like storage or backup are provided "as a service," meaning that they are available on demand, as needed, drawing on a pool of computing resources that from the user's point of view seems both unlimited and uncomplicated. The complexities don't actually go away, but they're hidden in the "cloud" of network services, so most of the time we don't have to worry about the mechanisms for provisioning applications any more than we do about the inner workings of a network router or an email server.

In other words, cloud computing is about providing the illusion of simplicity. It's about "making it look easy." When an IT department consumes cloud resources for its own purposes, or makes cloud services available to the users on the enterprise network, the service provider is expected to provide a clean interface and keep all the behind-the-scenes technical challenges to itself. But providing access to cloud services presents many of the same challenges as deploying applications on a corporate-wide area network (WAN). If anything, ensuring adequate performance for cloud services is more challenging for cloud services accessed over the public Internet. Network bandwidth is limited, and the transmission of signals is delayed when they have to be relayed over long distances.

Many enterprises are also constructing private clouds - virtualized pools of storage and processing resources and on-demand resources operated within the firewall that mimic the architectural style of public clouds. Just as more corporate users are accessing cloud resources over the Internet, more branch office employees are accessing private cloud resources over the WAN.

Fortunately, the challenges of cloud performance can be addressed with many of the same strategies IT organizations have employed to improve the delivery of applications and network services to branch offices, telecommuters, and mobile workers. WAN optimization technologies have evolved to address the issues of long-distance computing, and they work equally well over public and private networks.

Here are five reasons why WAN optimization will be even more important in this new era:

1. Cloud architectures, public and private, have a lot in common with the delivery of applications over corporate WANs. The fundamental obstacles to performance on a corporate WAN include distance and the limited available bandwidth, which often clash with the demand for network access from applications designed to function on a local area network. Some of the limits of performance on a WAN are absolute, because of constants like the speed of light. Yet dramatic improvements are possible with the WAN optimization appliances and software, which compress the data to be transmitted over any network link and streamline unnecessarily chatty application protocols. These techniques work just as well over a public Internet connection as over a private WAN circuit.

In a research report, "Increase IT Performance from the Enterprise to the Cloud with WAN Optimization," analysts from the Taneja Group depict current enterprise computing trends toward consolidation and mobility leading to "the emerging ‘cloud decade' in which IT performance will be constrained, more than ever by network performance."

2. Private clouds essentially continue the long-term trend started by server consolidation and virtualization. As these initiatives extend out beyond headquarters, more and more application services that were previously delivered by servers in each branch office are being centralized in corporate data centers. Delivering acceptable performance to these field offices will increasingly depend on WAN optimization.

3. Public cloud services rely on the open Internet to function as their WAN, as opposed to a more controlled network based on private telecommunications links. That makes achieving reliable high performance even more challenging, given that individual organizations must compete for bandwidth on the public Internet, where the potential for congestion is unpredictable. So making maximum use of whatever bandwidth is available at any given moment becomes doubly important.

With a public cloud service, headquarter users are in the same position as those remote office workers who have traditionally benefitted from WAN optimization. In other words, all employees will be remote from a data center operated by an external organization. How well the resulting performance issues are managed will be just as apparent to the CEO as to the most lowly branch office worker.

Even if your service provider and WAN optimization vendor don't have an explicit partnership to deploy optimization appliances, there are other ways to achieve the same effect. The optimization technology can also be delivered in software-only form, making it possible to deploy traffic acceleration as just another application running on a virtual container in the cloud infrastructure. While an appliance can take advantage of some hardware-specific functionality such as fail-to-wire capability, basic WAN optimization functionality can be delivered as software.

4. WAN optimization can accelerate both the public and private portions of a hybrid cloud architecture. Hybrid models using some combination of public and private resources are likely to prove the most attractive for many enterprises. Some data and applications will be kept within the firewall, for reasons of security and control, while other data and applications will be sent into the public cloud to save money and reduce complexity. In other cases, public cloud services will be used to provide overflow capacity or disaster recovery capabilities.

Even where cloud services are applied to background functions like data backup, WAN optimization will be important to minimize contention with other applications on the network. Moreover, when backup data needs to be restored, it often needs to be restored quickly and WAN optimization will help here.

5. The economic and architectural virtues of cloud computing have to be balanced with usability. Any interactive application that relies on cloud computing infrastructure needs to meet user expectations of reliability and performance. Those expectations will be formed partly by their experience with LAN-based applications (compared with which performance should not be noticeably worse) and with consumer Internet applications from companies like Google and Amazon.com, which invest heavily in maximizing the performance they deliver.

In an Enterprise Strategy Group technology brief ("Accelerating Cloud Performance with WAN Optimization," August 2010), principal analyst Jon Oltsik concluded that high performance will be a key success factor. "Private or public cloud ROI won't matter if users experience reduced productivity due to unacceptable response time to access files, applications, or their virtual desktops," he writes.

WAN optimization allows an enterprise to maximize the potential benefits of cloud computing, while minimizing a key tradeoff - concern over the performance of applications accessed at a distance. In the absence of a workable answer to this concern, organizations may not have the confidence to exploit the promise of cloud computing. Or they may go forward, but fail to deliver the performance required to prove the case for cloud computing within their companies.

Either way, failing to pair cloud computing with WAN optimization is a mistake. Only by making high performance part of the foundation of cloud computing can a company realize both the evolutionary and revolutionary benefits of this technology.