Dec 31 2009

Keeping Your Cool

As agencies converge communications over IP and adopt Power over Ethernet, IT shops have to focus on maintaining uptime and controlling heat.

Network convergence, led by the integration of voice and video into the data network, is changing how agencies use, manage and support their networks.

Unlike traditional private-branch exchange systems that serve individual offices or locations, Internet Protocol telephony distributes critical nodes and their power sources across multiple locations. The nodes support voice and data traffic, increasing the cost and impact of downtime across the network.

The density of new IP communications equipment — in data centers and wiring closets — is typically higher than that of the equipment it displaces. New IP switches cram more power into smaller packages, increasing power requirements and the heat generated. This is compounded when deploying Power over Ethernet (PoE) — powering devices across the same twisted-pair cabling that transmits data.

The current PoE standard, IEEE 802.3af, classifies devices into four groups based on power consumption (see Table 1 below). The Institute of Electronic and Electrical Engineers has a new standard in the works. PoE Plus, IEEE 802.3at, would raise maximum output power to 30 watts per port — almost double the current top end.

Together, IP telephony and PoE are driving the requirement for increasing uptime levels, power capacity and the need for dedicated precision air conditioning across the network. What this means is that agencies must factor power and cooling considerations into deployment plans to ensure their required availability levels. There won’t necessarily be a total increase in power consumption because of consolidation, but the individual equipment demands will require some changes.

For example, when specifying power and cooling for IP switches supporting PoE, agencies need to account for losses within the switch and from the switch to the powered device. Systems managers need to remember that electricity used by the switch is converted to heat, which can raise room temperatures to levels that can degrade equipment performance.

Uptime All the Time

Availability is the percentage of time a system is online and capable of productive work. It’s typically described as an annual percentage, or number of “nines.” “Five nines” translates into less than six minutes of downtime annually. This is often the objective for IP telephony applications to ensure that the availability and reliability of Voice over IP communications are not compromised. Determining the level of availability that is required for an application is related to decisions about a centralized or decentralized uninterruptible power supply (UPS) system configuration and architecture.

UPS systems provide three functions that affect availability. First, they provide immediate backup power in the event of an outage. When planning desired power availability for an IP communications switch, consider whether a generator is present and, if not, how long UPS batteries need to power the switch. If no generator is present, a minimum of two hours of battery runtime is recommended for IP telephony.

The second function is power conditioning. In most applications, the UPS removes sags, noise and other power quality problems that can damage equipment or degrade performance. The type of UPS used determines how power is conditioned. Online double-conversion and line-interactive are the most common types of alternating-current UPS systems used in network closets and equipment rooms. Online double-conversion systems provide the highest level of power conditioning.

Third, the UPS provides real-time remote monitoring and controlled shutdown of protected equipment. Ethernet-based communications systems allow for monitoring of remote power devices and provide real-time event notification and alarms. They also allow remote restart at the UPS. Power-monitoring software monitors UPS battery capacity and executes a shutdown of the switch when necessary to prevent damage and packet loss.

Proactive monitoring can help avoid undesirable situations associated with a down IP system. There are four basic power-protection architectures, each providing a different level of protection and availability depending on UPS configuration (see Table 2 below).

Cool and Dry

Failing to maintain a proper environment for critical electronics, such as switches, can result in reduced performance

and premature failure. For sensitive electronics, environmental control means more than comfort air conditioning. Precision cooling considerations include the following:

  • Temperature control: Sensitive electronics are generally designed for a stable environment of 72 degrees, plus or minus 2 degrees. As densities and PoE increase, the capacity of typical air conditioning systems is both insufficient and inefficient.
  • Humidity control: Electronics must be protected from both internal condensation (high humidity) and static electricity discharges (low humidity). Typical building air conditioning systems cannot deliver this protection.
  • Air volume: Sensitive electronics require greater air volumes than building air conditioning can provide. Because IP leads to higher densities, higher air flow is required. Usually, an office system will provide two air changes per hour. But a room filled with electronic equipment might require up to 30 changes per hour.
  • Air filtration: Dust and fiber that accumulate in systems can also make them less efficient and generate heat, so the air conditioning system will need to filter out contaminants.
  • Year-round operation: The design for a typical building air conditioning system operates normally for 10 hours per day, from spring to autumn. But for IP-centric systems, the systems cooling the electronics might need to run 24 x 7, at outside temperatures down to minus 30 degrees.
  • Fault tolerance: To support mission-critical services, environmental control systems for PoE setups likely will require more internal redundancy than the general office cooling systems.

Network convergence brings higher densities, capacities and availability requirements to agencies’ critical facilities. By evaluating systems based on their ability to deliver uptime, power quality and cooling required by IP telephony and PoE technology, the IT organization can help in the transition to expanded IP communications.


Table 1: Current PoE Power Limits

IEEE Class Maximum Output Maximum Input
at Power Source and Powered Device
Class 0 15.4 watts 0.44 to 12.95 watts
Class 1 4 watts 0.44 to 3.84 watts
Class 2 7 watts 3.84 to 6.49 watts
Class 3 15.4 watts 6.49 to 12.95 watts

IEEE 802.3af classifies devices based on power requirements.

Table 2: Uptime by Design

Tier Availability Downtime/Year Typical Setup
Tier 1

Three nines


8:46:00 Transient Voltage Surge Suppressor
Tier 2

Four nines


0:53:00 Single UPS
Tier 3

Five to six nines

(99.999% to


0:05:30 to 0:00:31.6 Redundant UPS or DC UPS (one active power path to critical equipment)
Tier 4

Seven to eight nines

(99.99999% to


0:00:03 to 0:00:00.3 Redundant UPS or DC UPS (two active power paths to critical equipment)

The design of the power system determines availability.