Data Gaps – The hidden pitfall of Data Center monitoring.
When your DCIM solution is deployed, and when every device that can be monitored IS monitored – there is a perception that monitoring has been maximized and that everything that can be done has been done – to avoid unexpected failures and downtime. That assumption is wrong.
Despite great strides in the industry, not all gear supports monitoring. Using legacy gear is a significant part of the issue, but also a choice. Rack PDUs are a good example. The cost difference between an in-rack PDU with monitoring and a glorified power strip can be significant. This is particularly true if multiplied by hundreds or thousands of racks. In addition, the location of the rack PDU near the bottom of the power chain implies less criticality and can lead to choices that reduce monitoring capability.
Whether your gaps come from legacy gear or choices – you should not accept gaps in your monitoring. A quality DCIM solution can solve this problem – not just mask it, but solve it. Configurable FLEXIBLE calculations can fill in the gaps with valid data extracted from other sources. Here are some approaches that solve this problem with meaningful values.
Sum of Children: This asset derives a device’s power load by summing the total power loads for all devices downstream (1st level) in the power chain. It depends on the accuracy of your model and having power values actively reflected in your power chain structure – but the resulting value is accurate. An example is a power strip that feeds five servers and a network switch. If the servers and switch are being monitored with accurate power loads in your power chain, you can sum them to see the total load on the power strip.
Sum of Points – Similar to Sum of Children, but only requires that the downstream device loads exist as points – they do not need to be fully realized in your power chain (for cases where servers are not fully modelled as assets). If the RAW point values for the downstream devices exist, you can sum those points without adding the servers to your power chain model.
From Parent – A device’s power load is derived from points on an upstream device that monitors loads. For example, a CT monitor on a breaker panel can show individual loads. You can access the load placed by the in-rack PDU based on the breaker feeding it. And if genuinely flexible, you can derive the total power, even if the CT monitor only returns current instead of power – you can calculate the total power using the current from the CT and voltage from the parent PDU or UPS.
Calculated by Phase: For more complex devices, you should be able to select points that reflect the current load on each phase and the voltage and mathematically calculate the total power load using industry-standard formulas for multiphase loads.
Once you have these derived values, you should be able to define constants that reflect the gear’s capacity and target loads as currents or total power. This allows you to track against physical capacity and target capacities – such as 40% for a failover device or 80% for a fully loaded device.
A flexible DCIM solution can do all this – and fill in data gaps. This eliminates your blind spots and lets you have complete power chain monitoring, even when your gear does not support it.
OpenData does this today. It’s one of many DCIM leadership modules that can work independently or together to give you an incredible view of your data center’s capabilities and potential.
You can reach us at firstname.lastname@example.org or 1-888.323.0066.