How to sink the total cost of ownership for intelligent rack PDUs

How to sink the total cost of ownership for intelligent rack PDUs

Get ready to save some serious $$$.

Intelligent power distribution units have clear advantages in the data center, including, but not limited to:

  • Real-time power metric gathering. 
  • Embedded ports for hot-swappable monitoring intelligence.
  • Remote switching of individual outlets. 
  • Improved short- and long-term load balancing. 

However, if not properly configured, these benefits will come at a cost to your networking infrastructure. Every network switch has a finite number of switch ports that are used to support internet-connected endpoints, such as a smart PDU. Historically, the requirement was one smart PDU per port, a dynamic that invariably increased total cost of ownership for PDUs.

In recent years, top-tier PDU manufacturers have implemented designs that can drastically reduce TCO while still supplying the core benefits associated with smart power. 

Save thousands of dollars with fault tolerant daisy chaining

"Up to 40 intelligent PDUs can be connected via Ethernet cable to one another in a series."

Say you spend $20,000 on a network switch that has 48 switch ports. By this logic, each port costs a little over $416. Redundancy best practice is that you have two PDU installations per rack for A and B supply feeds, and you balance the load between them. In theory, this would mean that each of those PDUs would require a switch port if you wanted it to be intelligent. Mathematically, that equates to a TCO increase of about $833 per rack. This is before the cost of cabling needed to bridge each PDU back to the switch, which could tack on a few hundred to a few thousand dollars depending on the size of your facility. 

This is where next-gen network daisy-chaining comes into the picture, which is made possible through Rapid Spanning Tree Protocol (more on that in just a bit). In this setup, up to 40 intelligent PDUs can be connected via Ethernet cable to one another in a series, or chain. This doesn't mean that if one PDU fails, it breaks the entire chain. Rather, the series of PDUs are connected to the switch using two ports. One cable connects through the end of the chain while the other connects at the beginning, facilitating bidirectional network connectivity.

Now let's look at what this means for the numbers using the same parameters: With daisy chaining, 40 PDUs use only two network ports (combined $833). Again, assuming two PDUs per rack, this means you can provide intelligent power for 20 racks for the same network cost it historically took to bestow intelligent power on one rack. Your per-rack networking TCO for smart power is now a little over $41. Scale this across your entire facility, and the savings easily add up into the thousands. 

Resource efficiency is about much more than saving electricity.There's more to resource efficiency than just saving electricity.

Understanding RSTP

This is where we loop back around to the aforementioned Rapid Spanning Tree Protocol (RSTP), which is the enabler of fault tolerance. In standard daisy chaining, you use only one switch port, but with a caveat. Should any PDU or links between them fail, everything downstream of the fault is knocked offline, which means you lose real-time intelligence gathering for several racks' worth of IT equipment. This is because the series extends outward in a single direction from the switch without any loop-around at the end of the chain. And, unfortunately, fixing that problem isn't as simple as just adding connections at the beginning and end of chain (as we mentioned above) – network loops using traditional protocols will cause "broadcast storms" that precipitate a failure. 

However, bi-directional network connectivity is possible with RSTP, which facilitates two layers of communication. Should one direction be blocked off for whatever reason, data traffic is automatically rerouted along a different path to and from the network switch. As a result, a series of PDUs could technically be broken right in the middle without losing network connectivity, and without disrupting the real-time flow of the rack PDU intelligence. 

With RSTP-enabled intelligent PDUs, you accomplish two critical functions simultaneously:

  1. Facilitate smarter power in the data center at a fraction of the TCO. 
  2. Achieve fault tolerance – and more specifically, exponentially reduce the possibility of a disruption in PDU connectivity.  

 The result: High-performance, durable and affordable power distribution intelligence in the data center.