Smart Racks: The Deep End of the Pool

Add Your Comments

Greg More is Senior Product Marketing Manager, Raritan Inc.

Greg-More_Raritan_smGREG MORE
Raritan

Many data centers have grown larger and more complex in recent years as the trend has been towards data center consolidation. With increasing size and complexity, there is an increased need to drive intelligence to the domain of IT devices.

A data center, whether a room or an entire building, is all about what is happening at the rack. Obviously, this is where the actual computing is taking place. It is also where the vast majority of the power is being consumed, or should be. It is the web of communications within the data center, within the organization and for communications with the rest of the world. If this kingdom of IT is not properly protected and maintained there can be very serious consequences.

An interesting analogy to consider is the fish tank. At its simplest, there is the goldfish bowl. As long as the water is kept clean and the right amount of fish food is supplied to the goldfish all is well. But graduating to a more complex freshwater tank with multiple types of fish requires maintenance of a more complex environment and ensuring that all the creatures will coexist. And finally, maintaining a saltwater aquarium requires a great deal of attention including careful monitoring of salinity, PH level and nitrate level, the presence of ammonia, skimming protein, etc.

Following the fish tank analogy, the right monitoring and metering at the rack can lead to some nifty data center improvements – right sizing the data center and just-in-time expansions to save on capital expenses of data center build outs, improved energy efficiency, improved IT productivity and utility, and better integration with cloud computing.

Come On In, the Temperature is Fine

There are many items to monitor and meter at the rack:  electrical power, temperature, humidity, airflow, air pressure, cooling capacity and network capacity. There is also the matter of tracking IT assets.

All the monitoring, metering and tracking areas identified above require an intelligent gateway to pass the information to software which can aggregate, analyze and report the findings. The gateway can be a stand-alone device like a rack controller or it can be incorporated into something else needed at the rack such as an intelligent inline meter or power distribution unit (iPDU) provided the embedded computer/processor is sufficiently scalable to handle the required tasks over time. The different types of data to be gathered, the desired polling frequency and the architecture of the system will determine the amount of data to be buffered. It is important to note that some data points are more volatile than others so polling frequencies may vary, e.g., current draw is more volatile than humidity. The idea is to ultimately fully instrument the rack with an open and interoperable system with sufficient capacity to handle a variety of data with room to grow over time.

The controller should be capable of being accessed through an Ethernet LAN port to a Web-based GUI as well as command line interface (CLI) and have an LCD display for use when working at the rack itself. These communications alternatives ensure that the smart rack can be managed from anywhere. Obviously, user authentication and authorization including permissions, LDAP/S and Active Directory and strong encryption and passwords are important for any system. SNMP is a useful and practical way to gather data from a variety of devices via their MIBs. In addition to Ethernet and serial connections, USB-A and USB-B ports are also useful such as logging to a USB stick or supporting devices such as a webcam.

Typical environment monitoring sensors include: temperature, humidity, air pressure, airflow and contact closure to detect things like open doors. Some sensor configurations make it easy to follow best practices in monitoring temperature and humidity such as ASHRAE guidelines to place temperature sensors at the top, middle and bottom on the cool air inlet side of a rack. With the right software, data points from the sensors can be plotted on a chart mapping temperature vs. humidity to determine if your IT equipment is operating efficiently and within its thermal boundaries.

Keep Your Head Above Water

By moving intelligence into IT equipment racks, data center managers can ensure that their critical IT infrastructure is healthy and happy.

Power metering, monitoring and control are the domain of intelligent PDUs at the rack and should be done at different levels within the rack:

  • At the overall rack PDU itself (single phase or three phase)
  • At circuit breakers within the PDU to know how close a breaker is to tripping
  • At each line for three-phase models in order to balance the lines
  • At individual outlets or user-defined groups of outlets, possibly across multiple PDUs such as when determining the power consumption of a device with redundant power supplies.

To understand actual power consumption requires kWh data, ideally at billing-grade levels of accuracy, e.g., +/- 1%. Other useful power measurements are current (amps), Watts, voltage, volt amps and power factor.

With a “smart rack” a number of energy and event management questions can be answered:

  • How can I save energy?
  • What is the total cost of the energy I use?
  • Who’s consuming energy and how can I bill them for that energy usage?
  • How do I maintain a safe environment for IT equipment?
  • Am I operating within ASHRAE temperature guidelines?
  • When do I inject humidity, bring in fresh air, mix hot and cold, run chillers?
  • How much power capacity do I have left for data center growth?
  • Can I get alerted when I’m loaded more than 40% on a feed?
  • How can I begin to manage CO2?

With a “smart rack” a number of asset, capacity and change management questions can be answered:

  • What do I have? How is it configured and connected?
  • Where is it located and who owns it?
  • What’s the maintenance on it?
  • How much space, power and network do I have available?
  • When do I run out of capacity?
  • Where can I put stuff?
  • How do I better utilize capacity?
  • How do I manage moves, adds, deletes? What is the impact of that change?
  • Who does the work? When is the work done? How do I know it’s done correctly?

Don’t Drink the Ocean

You don’t need to implement all the possible smart rack solutions at once. Ideally, you should be able to deploy what you need now, or what will provide the greatest improvement or fix the most pressing problem, without a major rip and replace of what exists. However, you want to keep your options open to implement additional improvements in the future without undoing or redoing past fixes.

It is possible to create a smart rack using the tools and technologies from a variety of vendors or from a single vendor’s comprehensive solution. Some vendors’ alternatives may be better suited to a particular need or deployment approach in which case selecting from a variety of vendors might be the best approach. But it is also worth considering a complete solution from one vendor to ensure interoperability and seamless integration. Also important is minimizing duplication which can occur when solutions from different vendors are merged.

Whether a single vendor or multiple vendors are chosen, remember that you don’t need to do everything at once. In fact, taking on too much can lead to paralysis as you try to decide what to do in what order.

Take the Plunge

Identify the area or areas that are most important for you to monitor, meter and manage now. Also, consider what might be next so that you are positioned to continue to make improvements. You do not want to implement a solution that will require significant modification or dismantlement to make additional progress on future projects.

Creating “smart racks” is a good way to implement data center infrastructure management (DCIM). A typical definition of a DCIM solution would include the following:

  • Asset management: Know what assets you have and where they are located
  • Capacity management: Understand the resources available and the constraints including physical connectivity, network paths, cooling, etc.
  • Change management: Manage moves, adds and deletes
  • Power management: Understand how much power is being consumed where
  • Environment management: Understand the conditions in the data center
  • Energy management: Understand what is consuming energy and how you can minimize the costs

Use technology at the rack to create “smart racks” which can make intelligent use of all your IT resources including IT assets, power, cooling, networks and, of course, people.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.

Add Your Comments

  • (will not be published)