During economically-challenged times, datacentre consolidation projects remain an attractive way of saving costs – but what are the critical challenges that must be faced to ensure success? Adopting an application or service-centric led strategy is key to successful consolidation programmes. Kevin Gunner, senior consultant at C&C Technology Consulting explains….

The modern datacentre is a very expensive and complex resource. With continued pressure on IT Managers to reduce costs, and deliver more, with less, without any degradation in service quality or availability, it comes as no surprise that datacentre consolidation activities continue to present challenges to the modern enterprise.

During years of growth, an organisation may expand beyond the traditional norm of owning 2 datacentres to perhaps acquiring 3 or 4, sometimes more server rooms that can be the cause of many problems often owing to their lower specification and suitability, whilst still incurring costs and significant risks through ownership. Now that we are living in leaner times, and virtualisation platforms and management tools are well-established, mature and reliable technologies, many of those companies are slimming down, looking to reduce the significant cost overhead of owning multiple facilities, by spreading the critical services over 2 main datacentres.

With cost savings and ROI as the primary driver behind such large projects, other critical aspects such as application performance and capacity are often overlooked or are not given due-consideration. Post-consolidation, this can have serious effects on end-user productivity, morale, brand value and ultimately, bottom-line revenues.

Take, for example, the subject of Application Performance. Once the datacentre move is completed, it is imperative that applications continue to deliver at least the same levels of end-user experience (if not better). Depending on the geographies or distance from the end-user community involved in the transition, coupled with the nature of the application in question and the extent to which it is sensitive to inherent WAN latencies, this is not always the case. The result? Unhappy users, frustrated that their critical line of business applications are too slow to use, or in extreme cases fail altogether. Dissatisfied customers, who lead busy lives themselves and don’t understand why it requires an unacceptably long time to complete a simple transaction. They may even consider taking their business elsewhere.

It is vital to ensure that your business applications will continue to work as designed after the consolidation. How? The answer to this question lies in profiling, modelling and prediction.  Prior to the consolidation, by understanding key performance characteristics of your candidate applications, and the network conditions imposed by it in terms of latency and available bandwidths, you will be in a position to make an informed judgement with regards to the likely impact on end-user experience.

But what about capacity? Obviously you’ll want to ensure that the links you provision into your consolidated datacentres can handle not only the current traffic load, but also the additional workload introduced by the consolidation activities. Again, profiling will deliver the answers, from the general traffic load over an extended period of time to the all-important per-user network footprint enabling you to confidently make provisioning decisions with your supplier.

Tools from many vendors can provide crucial support during the pre-consolidation profiling phase. For profiling and modelling or prediction, leaders in this field for example Opnet with their apptransaction-xpert offering or Compuware’s  Vantage for Transaction Profiling are well established and for full WAN simulation a product like Virtual Enterprise (VE) Suite from Shunra will  ensure that transaction response times are understood fully so that the proposed application deployment strategy can be modified accordingly.

During a recent engagement, which involved the enforced sale of an airport facility in the South East, the customer was required to transition all airport operational and business applications to a pre-existing datacentre facility that was reasonably close by. In this case, profiling was used to ‘sweep up’ applications and services still being accessed by users post separation, by gathering and classifying network traffic over an agreed period of time.

What can you do if you find that the planned consolidation isn’t going to work from a performance perspective? At the very least, this will buy you some time to consider alternate architectures for application delivery or to examine other strategies that can help improve end-user perception of performance. These include thin-client delivery mechanisms that vastly reduce the network overhead of client connections or via the deployment of WAN acceleration technologies, whereby frequently requested content can be cached to reduce WAN usage. Both of these techniques can improve end-user experience in situations where a previously local instance of an application is made remote by datacentre consolidation.

The important point to bear in mind is that end user experience is the key metric by which an application’s performance must be gauged after data centre consolidation. Your data centre, device and network monitoring may be showing normal levels of service and throughput. However, if your customers are abandoning online shopping carts and defecting to your competitors’ sites, or if they are getting frustrated by delays in getting through to your call centre, then this is a key indicator that your application performance has degraded to unacceptable levels. Undertaking application performance profiling and stress testing can help to avoid the common pitfalls of data centre consolidation, while still delivering the desired cost savings.