What’s the future for the datacentre?
The two themes of virtualisation and cloud computing are overlapping in the IT industry. At the heart of these trends is flexible and elastic computing power, which is still supported by the physical IT assets within the datacentre.
While virtualisation consultancy and cloud computing skills are in-demand with customers and providing valuable services revenues to the channel, the physical side is often overlooked. Many installations are outsourced as the channel considers there is little opportunity for margin. Solution providers taking this approach ignore some of the principle customer requirements for these deployments: guaranteed performance, minimum maintenance, predictable PUE (Power Usage Effectiveness) and risk mitigation.
While virtualisation and cloud computing projects are seen as delivering computing resources independent of the physical IT underneath it, the reality is that data centre design needs to be considered from a deployment perspective maximising airflow, reducing cooling costs, cable management and grounding and bonding. Planning for a given PUE has to be monitored and maintained in order to be successful. Even with the latest most efficient server technology, unified switching fabrics and virtualised storage the physical deployment holds the key to the system performance and delivering the virtualisation value.
Traditional datacentres were typically split into zones depending on the mission critical nature of the business application being supported. Non-mission critical applications are provisioned on lower cost hardware, minimum performance cabling and with little attention to resilience. While business critical applications are hosted on beefy servers, large storage arrays and cabling with the highest head room and maximum resilience. The datacentre would then be monitored for hot spots and special cooling provisions made for high temperature zones.
This segregation is not possible in a virtualised datacentre. With the implementation of virtual machines, applications are no longer tied to specific servers as they can be in constant motion around the data centre. The underlying resources are continually monitored for capacity and performance and matched to the required load in real time. Applications are being moved from one physical location to another physical location automatically as one resource is deemed more appropriate for the given applications load.
This means that any hot spot created would be moving around the datacentre following the application. This cannot be rectified after the fact, so the planning and deployment phase must ensure that the requirements for power and cooling resources are focused across the whole data centre, rather than just being a datacentre room solution. Secondly, utilisation of the computing resources has been driven up, leading to more consumption and greater density of assets. Again, this drives up the potential cost of power and cooling across the data centre. This is driving the trend for containment strategies, which offer up to 30 per cent reduced cooling OPEX over traditional Hot and Cold Isle build outs.
The mobility of applications across the datacentre also makes tracking of the correct physical resource hosting the application a challenge. A service desk receiving user complaints will need to identify the virtual machine running the application, map this to the real server hosting the virtual machine and then map this to a physical location to enable maintenance staff to function efficiently. This is driving a new set of datacentre management tools that combine deployment monitoring of IT equipment and the physical environment (power consumption, temperature, airflow, humidity) along with the physical network connectivity. These tools are known as Data Centre Infrastructure Management (DCIM) solutions. This creates new opportunities for the channel in consulting, deploying and maintaining these tools. Some will look to offer DCIM as a service.
