Paul Hollingsworth, product director for Celona Technologies says that by consolidating you can increase performance and save money.

Business managers are waking up to the high, and increasing costs, of running their data centres. A recent report from the Lawrence Berkeley National Laboratory (LBNL) in the US, for example, estimates that powering and cooling servers in data centres costs businesses an estimated $7.2 billion per year. The report warns that data centres are expected to increase their power demands by at least another 40% in the next three years. This situation is not unique to the US, however, and is reflected in figures released by the UK’s Department of Trade and Industry, which calculated in October 2006 that the UK’s computers and servers were now consuming more power than the entire state of Luxembourg.

The true costs of deploying applications or storing data have traditionally been poorly understood, with a focus on capex rather than on the lifetime cost. According to IDC’s John Humphreys, power and cooling and other operational costs account for 70% of a server’s lifetime cost. The rising cost of electricity, and the uncertainty over supply, has brought these figures into sharp focus for many businesses seeking to increase their efficiency for a variety of reasons. Reducing electricity consumption in data centres reduces opex, increasing profits or releasing cash that can be better spent elsewhere, and a positive side effect is that it helps to reduce a business’s carbon footprint relatively painlessly.

But electricity consumption is not the only reason why businesses are seeking to optimise their data centres’ performance. Real estate is expensive in a lot of key locations, and in some both the power supply and the available space for data centres is reaching capacity (for example, in financial centres such as the City of London). At the same time key new pieces of legislation are coming into play across various industries which require businesses to increase the amount of records they need to store – examples being the EU’s MiFID and Data Retention legislation. MiFID in particular has an aggressive deadline of 1 November 2007, which the vast majority of the financial services community will not meet. A short term constraint in achieving compliance is that many financial firms do not have the data centre capacity to host new applications that are being selected to aid compliance or to store the huge amounts of data that will need to be stored. While they have the option of building new data centres there is not sufficient time to do this before the 1 November deadline, so their remaining options are either to outsource or to try to make better use of the resources they already have.

The good news is that there is great scope for efficiency. In many businesses server utilisation is woefully inefficient, with a typical strategy being to deploy each new application on new servers. This is a result of departmental purchasing and the fact that IT directors have typically been targeted on metrics such as performance, rather than efficiency. In many large enterprises there are multiple redundant and duplicated applications, and many data silos holding duplicated data. Switching off applications and consolidating them liberates server space; as does reducing the amount of duplicated data. Moving from legacy to new generation applications can sometimes also increase efficiency, because some modern applications offer increased levels of functionality while also needing less server space.

Thus far it seems like a no-brainer: increasing data centre efficiency is good for everyone right? So why aren’t more enterprises doing it? Part of this answer we have already discussed: there has to be a driver towards efficiency. But as we have seen, multiple drivers are emerging across a range of industries. Another problem though is that consolidating data and applications is far from straightforward, relying on successful application data migrations. And herein lies the problem: data migrations of large and complex applications have traditionally been risky and costly, with a high probability of failure, leading CIOs to develop strategies to maintain their legacy systems.

Why are complex data migrations so problematic? Firstly they are highly complex, especially when there are large volumes of data to consolidate. And often insufficient resources are assigned to this part of the project at the right time, due to an underestimation of the challenges, assumptions that existing tools and process are sufficient, and an overestimation of data quality.

Typically, it is only when the project is well advanced that enterprises discover that they really do not understand their data. They find it is ‘dirty’, inconsistent, does not follow schema rules and is duplicated. This is because over many years the user community has bent and forced data into the system – often because the data structure was too inflexible or through lack of understanding. When moving this data into new applications it is not simply a case of lifting and shifting. The data has to be properly deconstructed, and reconstructed as expected by the new systems. This single issue causes application migrations to stop and start continually, meaning that overall they progress at a snail’s pace! And in this case time is the enemy, as the longer the migration takes the more likely it is that the transferred data becomes ‘stale’.

Traditional methodologies and tools (such as Extract-Transform-Load or ETL) have not really solved the problems of complex data migration, as they were designed for another purpose (bulk load of data warehouses). Phil Dance, BT’s Technology CIO, comments: "For large-scale mission-critical applications ETL doesn't work as a conversion strategy…However compelling the new application, the function without the data is meaningless."

Fortunately, a new generation of application data migration tools are now available that are designed specifically for complex application-level data migrations. These tools use an approach known as "progressive migration". The progressive approach is highly flexible: it does not cement the transformation process in place and allows transformation momentum to be built up and maintained. The approach and toolset support real-time synchronisation of data between the old and new worlds (bi-directionally if required), ensuring that the business transformation can be completed independently of the technical transformation. Using third-generation tools means that businesses can more easily consolidate their IT infrastructure, helping them to make far more efficient use of their current data centre capacity.