Published: 21 Nov 2011Date: January 18, 2012
IT organizations are increasingly being called upon to cost-effectively deliver reliable support for the entire catalog of business services, or risk outsourcing to a managed service provider. Previously, IT architects would use historical trends to predict capacity requirements and simply over-provision to account for any peaks caused by seasonality, error, or extraneous influences like mergers and acquisitions. Over-provisioning, combined with poor lifecycle management of new resources provisioned in the data center, has led to capacity utilization and inefficiency issues.
While historical data is great for understanding past issues and current state of the environment, the performance of servers, hosts and clusters is not linear; at some level of saturation, the performance of that infrastructure will quickly start to degrade. The impact is that business services dependent on that infrastructure suffer, and users experience longer response times, unavailable applications and unacceptable performance. Lack of predictability and uncertainty has caused many IT organizations to suffer “VM Stall,” meaning they have been unable to advance to a strategic deployment of virtualization and achieve expected levels of consolidation. IT must understand the impact of change on the infrastructure to ensure performance. Without that insight, IT is unwilling to take that risk.
To proactively plan for capacity in order to support mission critical business services, IT must assess historical data and couple it with predictive analytics. This allows you to efficiently manage what the actual capacity utilization of that business service is and what it will be as the workload grows, the infrastructure virtualizes, or other changes are introduced. Given this predictive capability, IT could effectively foresee capacity issues in the future to assess alternatives and provision the right infrastructure when needed.