Shifting Workloads to the Cloud? Not as Easy as It Sounds

The cloud may be the most advanced information architecture to date, but in the end its value can only be determined by its ability to handle data workloads.

Initially at least, many organizations are looking to the cloud as a means to expand their storage capabilities. However, this only scratches the surface. Once architectures have progressed to the point where loads can be shifted across dynamic pools of internal and external resources, enterprises will be able to drive levels of efficiency that were unimaginable only a few years ago.

The rudiments of such an architecture are already coming into place. Fujitsu, for example, recently showed off a prototype of a new cloud platform that it says allows unprecedented flexibility in configuring resources to match data requirements. The package does away with application-specific servers in favor of a high-speed interconnect architecture that mixes and matches CPU and storage resources at a moment’s notice. The company has also developed a new management/middleware stack specifically designed for resource pooling and seamless back-end configuration management.

To get the kind of workload flexibility the cloud is capable of will require a high degree of coordination among disparate resources, according to Red Hat’s Gordon Haff. That means you’ll need to be able to cross multiple virtualization stacks and cloud platforms using a single management platform. In that light, your cloud needs to be able to identify suitable targets for your data and then preparing workloads for each environment. And it has to do this in a way that is both transparent to the user and maintains a high degree of consistency across various resources. Fortunately, Red Hat has just such a product in beta.

Still, the most robust automation in the world won’t amount to much if you don’t have the right governance policies in place, according to ServiceMesh Inc.’s Derick Townsend. Resource relationships in the cloud exceed the complexity of those found in virtual environments that are already giving enterprises managers fits. Cloud governance done wrong will ultimately hamper your ability to provide things like immediate self-service, automatic configuration and scaling and workload optimization. A properly governed cloud will include a pre-configured zoned security model, as well as data regulatory mechanisms that account for new geographic and distribution parameters.

Many organizations also might discover that some workloads simply are not suitable for the cloud. As researchers at Penn State University found in a recent analysis, large jobs may actually prove too unwieldy for the cloud, ultimately costing more than if handled strictly in-house. You also have to take into consideration the popularity of the application and what its licensing structure looks like. Data transfer and migration costs can also mount over time.

All of this should put paid to the notion that the cloud is simply a giant plug-and-play data center. True, it places a tremendous amount of resources at your disposal, but it also presents a significant management and integration challenge, especially if you intend to leverage it to its full potential.

That doesn’t mean you need to rethink your cloud plans, but you should take a hard look at its current and future capabilities before you find yourself relying on it too heavily.

Latest Articles

Follow Us On Social Media

Explore More