Network engineers play a critical role in the design, upkeep, and smooth operation of the networks they manage. A key aspect of this responsibility is ensuring that network equipment receives reliable, well-managed power to maintain redundancy, ensure high availability, and meet capacity demands. Proper power management supports network performance and helps prevent costly downtime and equipment failures.

In this article, we explore the essential principles of network power management, offering best practices and strategies to help engineers optimize power efficiency while ensuring the resilience and reliability of their networks.

The role of power in network management

Power is an often-overlooked part of the network administrator's responsibilities, usually due to the fact that it is considered a separate discipline. However, it is a tightly associated aspect of network design, and as such, network administrators are obligated to be knowledgeable in this area to ensure power optimization for their networks.

There are three primary aspects of power management that administrators need to address:

• Power redundancy at telecom closets and data centers
  
  
• Consolidation of power using PoE
  
  
• Backup power provided by UPS systems and generators

The importance of power redundancy in power management 

Enterprise network design includes the planning and layout of data centers and telecom closets. These are the physical locations where network devices such as routers and switches are installed and interconnected.

Telecom closets are dedicated rooms or spaces in a building that house networking and communications equipment. Telecom closets contain the intermediate distribution frame (IDF) used to distribute cabling to the end user, reaching desktop computers, IP phones, and wireless access points.

Typically, each floor, wing, or section of a building will have a telecom closet where all of the cabling for that area is concentrated to be connected to a network switch.

Switches in telecom closets must be provided with redundant power. This is done using several techniques:

• Dual power supplies: Higher-end switches have dual or multiple power supplies. Switches can function with only a single power supply, ensuring that in the event of a failure, at least one supply will function and allow the switch to continue to operate. Power supplies are often hot-swappable, which means you can replace them with zero downtime.
  
  
• Redundant power provisioning: The grid power that is supplied to the telecom closet should come from multiple power circuits connected to different circuit breakers. This way, if one circuit breaker trips or a power circuit fails, the second power circuit will continue to operate, ensuring power availability to devices.
  
  
• Uninterruptible power source (UPS): At least one circuit (preferably all circuits) serving the telecom closet should deliver uninterruptable power from a UPS/generator combination. This aspect is further discussed later in the article.

A data center is a facility that contains computer systems and related components—including servers, storage, and networking equipment—that store, process, and distribute vast amounts of data.

While power management concepts are similar to those applicable to telecom closets, data centers demand significantly greater redundancy and scale in power management.

Thus, all of the principles involved with telecom closets also apply to data centers. In addition, here are some of the most important power management principles that apply to data centers in particular:

• Per-rack power delivery: Typically, each rack should have at least two separate circuits delivering power to the devices installed on that rack. Depending upon the power budget of the equipment on that rack, it may be necessary to provide additional circuits. An independent circuit breaker should serve each circuit.
  
  
• Uninterruptible power: Power arrangements for the data center should be such that all circuits providing power to all racks are protected by uninterruptible power supply (UPS) systems supported by a diesel generator.
  
  
• Cooling system redundancy: A commonly disregarded aspect of power involves ensuring that data center cooling systems have the appropriate backup power in case of a power failure. Without at least a partial cooling system power backup in place, a data center could easily overheat in minutes, causing extensive damage and costly network service outages.

Taking advantage of PoE

Power over Ethernet (PoE) is an essential part of any network power management strategy. PoE delivers power to various access layer devices, such as IP phones, wireless access points (WAPs), videoconferencing endpoints, point of sale (PoS) devices, and IP cameras.

Network administrators should aim to use PoE whenever possible to power such devices. This is done by ensuring that access layer switches delivering network connectivity also support the PoE requirements of these devices.

When doing so, it is important to ensure that the power budget delivered by such a switch is sufficient for all devices connecting to it.

A PoE switch will typically be able to support only a certain wattage. Even if all of its ports support PoE, this does not mean the switch can provide the required power for all connected devices. For this reason, careful planning of the power budget of each switch is essential.

There are several advantages to using PoE as much as possible:

• No grid power needs to be provisioned at the location of PoE-powered devices. This can result in substantial savings given that the number of IP phones alone in a large enterprise may be in the hundreds or even thousands. Also, IP cameras and WAPs are typically mounted on walls or ceilings indoors and outdoors, and PoE eliminates the need to run power to these hard-to-reach areas.
  
  
• Centralized power redundancy can easily be provisioned for all PoE-powered devices by simply providing redundant power to the PoE switch serving those devices. Ensuring that telecom closets and data centers have redundant power enables redundancy for all PoE-connected devices.

Backup power systems

It's important to fit telecom closets and data centers with specialized electrical systems powered by high-capacity UPS systems supported by a generator. The following diagram shows the typical connectivity for such an arrangement.

Under normal circumstances, the power company provides power to the UPS system, which delivers power to the data center, telecom closets, and other vital areas for network operation. The generator remains on standby.

This normal operation mode ensures that network equipment is isolated from any harmful phenomena from the power company, such as spikes or brownouts.

In the event of a power failure, the UPS will continue to deliver uninterrupted power to the required systems until the generator powers up.

A specialized power switch that detects the power failures handles the coordination of the mechanism and intelligently decides if and when to power on the generator and switch power input to the UPS. It is also responsible for switching back upon power restoration from the power company.

Some important aspects of this arrangement include the following:

• In the event of a power failure, the generator should not start immediately because many power outages may only last for several seconds, resulting in an unnecessary engine start. For this reason, one should configure a threshold of 30 seconds to two minutes before the generator powers up.
  
  
• During this threshold, the UPS provides all the power to necessary systems.
  
  
• The switch will go back to grid power when it is restored. This should also have a threshold timer of 30 seconds or more to avoid frequent switching.
  
  
• UPS capacity should be capable of maintaining power for several multiples of the start-time delay.
  
  
• Generators and UPS batteries should be tested monthly or bimonthly by going through industry-standard checklists to ensure peak operating conditions.
  
  
• Power budgets for UPS systems and generators should consider cooling systems vital for data center and telecom closet operation.

Conclusion

Electrical power may not be the first thing that comes to mind when considering network administration, but it's a critical aspect of managing enterprise networks.

While administrators don't need to master every technical detail of power systems, ensuring network reliability and performance requires a solid understanding of power budgets, the operation of UPS and generator systems, and the specifics of PoE implementation.

You may also like: 

Essential steps for applying QoS on UC networks

Nine DHCP options that are particularly useful for VoIP and UC

Unlock the incredible power of IPv6 for VoIP & UC systems