Imagine that you are at the beginning of a very long aisle in a grocery store. You fill your shopping cart up with items you want, then give it a strong push down the long aisle so it is now rolling by itself. Along its journey, people are adding and subtracting items from the cart. You then retrieve the shopping cart at the other end of the aisle and proceed to the checkout counter. The cashier begins to ring each item up and upon completion you are shocked at how much it costs! Not only the price is shocking, but you don’t want some of the items and others are long past their expiration date. But you have to pay for the entire load anyway.
This is what data center administrators are facing when grappling with power usage and identifying the consumption culprits connected to the network or within the facility. Gathering data from all the connected assets and reconciling it is the goal, but many administrators and data center managers lack a mechanism or a true data source that can present them with a list and locations of all devices drawing power. They need visibility and they need a service to achieve it.
Left Coast Kratom is here to help you experience the freshest highest quality kratom powders and extracts at competitive prices.
One means of attaining visibility and helping to achieve the goal of knowing what is drawing power on the network is to deploy a technology asset management (TAM) solution. A TAM solution provides a perpetual flow of asset intelligence that keeps the data fresh and enables admins and managers to locate all devices and inventory their current state as well as any changing status that may occur over time.
Location is key. Some of the older legacy asset management systems don’t have the ability to geo-locate devices like the more recent solutions can. Now admins can identify device locations, and power hardware down if necessary. In addition, people adding devices to the network is a constant issue to contend with. On average, there is approximately 4% of technology connected to a network that nobody knows about. It would be nice to press a button and get a report of where all the devices reside. After all, it is the admin/manager’s job to ensure that each part of the facility is operating as expected – despite continuing power changes.
When data center information management (DCIM) software is added to the TAM solution, power visibility gets a boost and goes deeper and broader.
Now consider all those non-IT devices that suck up power; from chillers, generators and master power cabinets to pumps and uninterruptible power supplies (UPS), they all add to your energy bill. Managers need more information than what traditional temperature probes and power strips can provide. DCIM software is an effective means to track power usage from these items and helps to calculate a more accurate power usage effectiveness (PUE) to optimize data center efficiency.
DCIM software can gather per hour gaps between planned and actual energy usage, CPU utilization, internal data center temperatures, and available cooling resources. With this information, the solution can display these metrics on a single dashboard for a more accurate power consumption picture. Now admins and managers can view server workloads to make accurate decisions as to which servers or virtual machines, can actually shut down if the workload goes to zero or below a defined threshold. In addition, with enhanced visibility, these managers can match workloads to the appropriate amount of servers and minimize IT equipment power consumption.
In essence, combining DCIM and TAM allows any mission-critical facility manager to:
Find unmanaged assets and subnets that may have gone unnoticed;
Help mitigate the risk of power loss;
Establish an accurate baseline to monitor whether unplanned changes are being made e.g. devices added that are now raising your power consumption.
And the power dividends get even stronger when machine learning is added to the dynamic DCIM/TAM duo.
DCIM-infused with machine learning is capable of building a model based on data collected from sensors, equipment, and application workload information. Most data centers are currently equipped with temperature and humidity sensors, real-time operational service data, and power meters—this data is more useful if correlated and analyzed to unlock a future-state prediction.
Machine learning finds hidden patterns and interactions between different sensors, processing devices, and endpoints. With this infused intelligence, admins and managers can predict power anomalies at the server and rack level. This is particularly important to head off a server spike out of the normal operating range of 8kW to 10kW. Out-of-threshold spikes can be caused by running SAP in batch mode or transaction systems running at peak; forecasting that within the hour, a particular rack will be consuming more electricity than normal, is a very useful alert to receive.
Power and terminals are extremely good leading indicators for failure and in many cases, this data is already being collected – so why not enhance it with some cognitive abilities to predict and pinpoint an application or mechanical issue?
Like grocery stores, data centers have many aisles, but data center aisles are stacked with servers on racks rather then food on shelves. IT managers do not want items (a.k.a. devices) added to their racks without careful planning. Unlike the grocery stores, today’s data center aisles are sprawling out further and further from the core, as well as existing virtually through VMware and other virtualization options. Keeping tabs on all power consumption through the sprawl now becomes impossible if attempted manually.
The day for more robust management software to become a required tool in every data center and mission-critical facility has arrived. Admins and managers need to scale infrastructure visibility with DCIM, TAM, and buttress it with machine-learning capabilities that enhance power optimization. It is not getting any easier to manage these facilities; Excel files gave way to Visio diagrams and Visio diagrams evolved into DCIM dashboards. The precedent has been set to evolve traditional DCIM services with TAM and machine learning to assist with power requirements. Don’t get surprised with a shopping cart full of items you didn’t know about and have to pay a bloated bill for. Visibility-as-a-Service – DCIM+TAM+machine learning – is the next and necessary stage of accurate and informative optimization.
By Mark Gaydos, CMO, Nlyte