Raise utilization to lower costs
By raising utilization, while meeting service level and ensuring service quality, you can mitigate servers over-provisioning, optimizing data center operations, lowering energy OpEx, servers CapEx and facilities CapEx.
Server utilization in data centers have been found to be low, ranging from below 10% to approximately 35%, according to a New York Times article. For a virtualized data center, low utilization can be summarized to be due to these 2 problems:
- Initial Placement Problem
- Rebalancing Problem
1. Initial Placement Problem
This problem is due to workload not being optimally placed. When starting VMs, they need to be optimally and rapidly placed within a large dynamically changing facility. But this is non-trivial computationally and the amount of data required is immense, further compounded with size in a large data center.
2. Rebalancing Problem
This problem is due to servers in the data center not being rebalanced despite workload variation. Along with changing VMs workload (and workload can change a lot!), the VMs need to be rebalanced across a large data center, while meeting workload requirements, and right-sizing the number of servers to match workload requirements. This is also very difficult to accomplish computationally, involving large amounts of data, and made harder with increase in data center size.
Solving two halves of the problem
The Initial Placement Problem and Rebalancing Problem are two halves of the core of the low utilization problem.
By solving the Initial Placement Problem and the Rebalancing Problem, service level requirement can be met without having to over-provision servers, and service quality degradation can be avoided. Data center operations can also be optimized, lowering energy OpEx, servers CapEx and facilities CapEx.
Find out more about what we can do for you.
Get in contact