By Keith Budge

CLOUD computing continues to evolve as one of the dominant themes in the ICT industry in 2010. Businesses are struggling with rising IT costs, exploding data volumes, and the ever- evolving competitive challenges in increasingly dynamic market conditions. This has catalysed businesses and vendors to explore new options to maximise resources and minimise expenditure.

From an enterprise CIO's perspective, the cloud offers new, more efficient and flexible ways of sourcing ICT capabilities. Whether cloud services are enterprise-grade yet, however, remains a topic of debate.

The hype around cloud computing is about its scalable and 'elastic' capabilities and the belief that existing IT architectures and processes can be simply replaced by the cloud. While this is true to an extent, the reality is not as simple - in particular when it comes to cloud adoption - especially for large enterprises. Apart from these challenges, what are the options available and how can CIOs make the right choices?

Understanding the market environment

The demographics are significantly different. The US is experiencing massive data growth but this will be trivial compared to the volumes that will be generated as Asia's massive private and business populations come online.

Data flows in trickles and torrents across our businesses - as transactions, event streams, logs, e-mails, and in countless other forms. Data volumes are exploding - to the tune of 1.5x to 2.5x a year, and new sources and uses for data appear every day. These data sources compound and catch many companies by surprise.

Organisations are realising that their data volumes have far exceeded the capabilities of mainstream OLTP-oriented database systems. While these systems can comfortably handle low numbers of TBs (terabytes) and modestly complex analytics queries, they start to break down or require extraordinary tuning in the low 10s of TBs.

In market reality, many companies today need to store and analyse 10s or 100s of TBs, or even low numbers of PBs (petabytes). The technology to support these larger databases exists today, in the form of MPP (Massively Parallel Processing) Share-Nothing DBMS systems.

These systems are optimised for the processing required by data warehousing and business intelligence, and utilise 10s or 100s of nodes working in parallel with data automatically partitioned across the nodes to support fully parallel query processing. Simply put - what was once impossible is now possible, and at a reasonable cost.

The next equally important challenge that has remained largely unaddressed until now is the hundreds or thousands of data silos in organisations. A company might have one or more 'enterprise data warehouses', hundreds of departmental silos, data spread out in Excel spreadsheets and Access databases, and countless custom applications.

This all-too-common story reflects the organic evolution of these systems, and inherent tensions between those responsible for the operation of these systems (usually, but not always, IT) and those that want to use these systems and the data within them (usually analysts within business units). IT looks at this chaos and sees cost and complexity; while, on the other hand, business analysts look at these as barriers to getting their jobs done. Each group has perfectly reasonable demands that are inherently in conflict.

There is a new way of looking at this complexity - centralisation of data and 'self service' for end users. Each stakeholder gets what they need - IT provides a stable, secure platform that allows business analysts flexibility through self service.

Cross-border issues

The cross-border issues actually bring up challenges and opportunities when it comes to the strategic questions. If we can do computing anywhere, we can think about placing different parts of our organisation in different geographic locations. We have a lot more flexibility.

So our clients have a lot more flexibility to cross international boundaries with different parts of their operations. That of course has not only HR and taxation implications, but infrastructure optimisation issues as well.

A company could very easily double, then quadruple and then multiply even further the number of partner relationships they have just for their computing and their applications. And IT managers often have to manage these relationships and maintain service delivery levels. So there's this incredibly complex set of questions to be answered around how to implement a new paradigm such as cloud computing.

Greenplum harnesses the power of commodity hardware and open-source software. This ensures the best possible price performance while creating a flexible business and technology architecture.

While cloud computing is not simple, it is possible, and is already in use at some of the most forward-thinking companies around the world. Vendors like Greenplum offer unlimited opportunities to enterprises offering extreme price performance, massive scalability with 10s of TBs, all the way up to the largest known database in the world - eBay's 6.5PB (6,500TB) that runs on a single Greenplum Database cluster.

Nobody knows exactly how cloud computing will evolve but a few things are certain, data is growing, businesses need more value from this data and vendors are focusing in a big way to lead this change.