Preparing For The Cloud? Fix The Data Problem First

Cloud computing is rapidly becoming the holy grail of enterprise computing, with most CIOs, network managers and IT departments investigating how they can develop and leverage a cloud strategy. While cloud computing can often mean different things to different people, true cloud computing provides massively scalable IT capabilities as a service over the Internet, and can offer companies a single point of access to manage and meet their computing needs. When implemented correctly, cloud computing can help companies cut IT and personnel costs, produce a leaner IT environment and create a dynamically scalable infrastructure that lends itself to rapid growth.

In an ideal scenario, cloud computing can reduce the total-cost-of-ownership for a company’s IT infrastructure by moving resources to the cloud and minimising the physical infrastructure. It’s an intensely supported goal – in fact, Gartner named cloud computing one of the Top 10 Strategic Technologies for 2009, asserting that it can help companies deliver services in a ‘highly scalable and elastic fashion’. Despite at first seeming like an elusive buzzword, cloud computing is gradually beginning to demonstrate some real benefits that validate its longevity as an infrastructure strategy.

Unfortunately, many organisations put the cart before the horse by immediately attempting to use the cloud to minimise compute resources and consolidate their IT infrastructures. While the benefits of the cloud can be very real, when companies develop initiatives that focus purely on a compute strategy, they ignore a critical component that could significantly enhance the benefits received by shifting to the cloud. To paraphrase a famous United States presidential campaign – it’s the data, stupid.

Companies need a comprehensive data strategy that can complement, catalyse and drive their compute strategy. Most enterprises today maintain enormous distributed architectures across multiple, disparate data centres that require huge amounts of compute power to move the correct operational data to the right applications and people in real-time.

While moving resources to a private or public cloud can reap major cost-saving rewards, that move has to incorporate an elastic data infrastructure, one that makes data ubiquitous and available that much faster to the applications and users that need it the most. An elastic data fabric can also help reduce physical resources and lower costs associated with powering the data centre. It creates a more resilient data infrastructure that increases failure tolerance, reduces latency and delivers data more reliably, enabling companies to analyse and make better business decisions based on that data.

This requires an entirely new way of thinking about data management. In a traditional environment, managers are assured that applications will perform the way they expect them to, as database servers are running to provide them with the data they need to perform. But the cloud environment changes the paradigm, as resources are shared and virtual. Now thousands of disparate systems and applications can have access to the same data in real-time and the data infrastructure can dynamically scale to any size without investment in additional hardware or licences.

To achieve this, managers need to develop strategies that take into account the massive amounts of data maintained in their infrastructures. They also must learn to effectively manage, balance and deliver the data so the infrastructure itself becomes data-aware and elastic. This type of elastic data infrastructure will power dynamically scalable architectures that can add new applications or users with no visible drain on infrastructure resources.

A data-aware infrastructure is one that incorporates data at every step of the process and manages it via load-balancing, disk caching and other approaches to decrease latency and provision resources to balance increasingly large data stores. It includes the data elasticity required to manage dynamically changing workloads across globally dispersed data centres. And it provides highly available data to users and applications to ensure they can experience the same response time in a cloud environment that they have in their current model with local resources.

In a real-word scenario, a data-aware infrastructure provides the ability for users and applications to receive real-time updates as data they are working with is changed by another user or application. This removes problems that appear when users are working with data that has been changed since they started looking at it, and creates the ability to deliver applications where data dependencies are immediately reflected when the underlying data changes. Companies that use data-aware infrastructures are able to build, for example, such things as spreadsheets that change as the data in the underlying systems changes.

But where do companies start when seeking to implement an elastic data infrastructure to increase resiliency and manage data in preparation for a move to either a public or private cloud? Companies must ensure that there is a new layer in enterprise architectures that is focused purely on managing real-time operational data in the infrastructure. This layer should ideally allow applications to pull from the most frequently accessed data either directly within the database or from a single network hop away. It should combine the filtering semantics of messaging with active caching and an event framework that allows events to be delivered to the edges of the enterprise. And this layer should have the ability to query distributed data, gain access to it in real-time and use it to empower users to make decisions that drive business growth.

The rewards here can be tremendous and immediate. For example, we’ve run tests using Amazon’s EC2 cloud technology using both a data-aware and a non-data-aware environment. In the data-aware tests, results were two to three times faster than the non-data-aware tests. That means the data was available that much faster to the applications and the users who needed it. The implications of this are enormous from a resource perspective. Companies can share applications and data two to three times faster and dynamically scale to meet as many users and applications as required. It’s also important from a cost-saving perspective, as it translates into physical systems using two to three times less power (a major benefit promised by the cloud) and requiring two to three times fewer compute resources than initially used.

Sharing data in a cloud can create the opportunity for a new level of data awareness across the enterprise, as today changes in data in one system are often not immediately seen by other systems. The financial services industry was one of the first to adopt this type of innovative data strategy, and it is now being adopted by other data-intensive industries such as telecommunications, pharmaceuticals and even the government. These organisations realise that a cloud computing initiative that takes into account both a data and compute strategy can provide real-time access to reliable, resilient operational data that enables faster decision-making and can deliver true business benefits. This new way of thinking about data management will give users and applications access to data they wouldn’t have had previously, ultimately making for a faster, leaner organisation poised to achieve new levels of agility, growth and success.

Lazar is director of Federal Technology, GemStone Systems.