The biggest misconception and pitfall is to manage and provision data center infrastructure and the applications which run on them independently. Given the scale, complexity and required optimization of today’s computing infrastructure, organizations cannot afford to generously over-provision their facilities. This has a direct cost impact to the business.
Furthermore, applications and facilities are becoming evermore entwined: Critical infrastructure works best when it is aware of the workloads running atop it, and applications work best when they are aware of the infrastructure upon which they run.
There are numerous examples, including:
- Using predicted infrastructure failure to pre-emptively move workloads
- Managing a facility (server utilization, thermal set points, etc.) based on actual workloads, so facilities are not running full tilt while applications sit idle
- Using “end-to-end” information, from facilities to users, to detect application and security anomalies
There are several business impacts related to the above: cost of wasted energy, reduced application availability with revenue/SLA/customer satisfaction impact, and reduced agility due to the lack of flexibility regarding intelligent workload placement.