Power Protection for Modern Data Centers and Server Farms
Dec 24
Posted: under Information Technology.
Tags: High Density, Server Farm, System Resilience
Robin Koffler asked:
Data Centre managers have a lot on their plates. Pressured into squeezing more equipment into increasingly overcrowded spaces, energy-efficiency is also high on the agenda. Finance departments demand that costs are cut but that systems are kept up and running. In the face of a growing number of power cuts, it’s a wonder they can sleep at night!
Installing an uninterruptible power supply (UPS) need not add to the nightmare, as long as a few preliminary steps are taken.
What managers want from their UPS is: maximum uptime, system resilience and efficiency, low running costs and energy consumption, high density and a small footprint. How do they get that? This article tries to answer that question.
The Primary functions of a UPS. An uninterruptible power supply, primarily, provides a valuable source of back-up power in the event of a mains failure. It allows businesses to ride through short-duration interruptions; perform a controlled shutdown of non-critical systems and power loads whilst a back-up generator starts up.
A lesser-known function of UPS (but increasingly important one in data centre and server farm environments) is power conditioning. The voltage of raw mains energy can fluctuate wildly, causing power problems that are harmful to electrical equipment, particularly servers. UPS monitor and regulate mains power supply, keeping it to within acceptable levels and negating these harmful problems, thus saving on costly replacement or repair and data loss or corruption.
Maximising Uptime.Maximum uptime is achieved, firstly, by sizing loads and UPS correctly. This involves determining load power requirements, which can be confusing. On equipment labels and in user manuals, hardware manufacturers rate power usage unrealistically high. It is not uncommon to see loads running at only 50% of this rating. In addition, ratings given may be in Amps or Watts. Depending on the type of load, there can also be a significant difference between in-rush (start-up) and running power.
It is important, therefore, when sizing UPS to carefully consider: apparent power (VA), active power (Watts) and Power Factor (pf). The higher the input power factor, the more efficient the UPS will be. Today’s standard is upwards of 96%. Matching the right UPS topology and design to the specific application is also important, so too ensuring resilience and no single-point-of-failure. Additional runtime and/or some form of alternative power generation (a diesel generator, for example) and ensuring the whole power protection solution is consistently and properly maintained will add to the likelihood of 100% uptime.
Building-in System Resilience. System resilience (fault tolerance) comes largely from the UPS design and topology and how the system is configured. The primary components of resilience are the automatic static transfer switch that monitors the output waveform of the uninterruptible power supply’s inverter. Should it fail, due to a short-circuit, overload or fault condition, the switch transfers the load to mains supply without a break and automatically transfers it back once the fault has cleared.
Other levels of resilience come from whatever source of back-up power (batteries, diesel generator and/or a dual input power supply) is chosen.
Extended Runtime. When extending the runtime of an uninterruptible power supply, special consideration must be given to load category and type. Load category determines how critical it is to the continuity of business. Critical and essential loads need extended runtime and some form of redundancy. Non-essential loads (canteen facilities or printing services, for example) may need a solution to power emergency lighting (for health and safety whilst they are shutdown for the duration of the power cut) but no extended runtime. Runtime is load dependent: the bigger the load, the more power it will need.
Load ‘type’ refers to whether it is linear or non-linear, and how it draws its current from the mains power supply. It will be inductive, capacitive or resistive and this will have a bearing on UPS sizing.
Low running costs and energy consumption. Leading UPS manufacturers, such as Riello UPS, have made great strides in recent years towards providing the market with energy-efficient products. UPS energy-efficiency depends on: the energy they need to power them; the way they draw that energy; their affect on upstream supplies and the management of consumables (such as battery sets and switchgear). Modern UPS designs offer tangible improvements in all these areas, along with an ever-improving carbon footprints.
The latest zero-impact designs, such as Riello’s Multi Plus, demonstrate low input current distortion, reducing the need for dramatically oversized supply (generators and upstream switching). They give a near Unity power factor sinewave, leading to less wasted energy and are physically smaller, thus allowing more revenue-earning, high-density servers to be fitted onto floor space previously required for UPS and batteries.
Plan for the future – today! Data centres and server farms are fast-moving environments where growth and expansion are constant. When choosing a UPS solution for today, thinking about tomorrow (specifically expansion plans) is just as important. The most common approach is to ramp-up load sizing by 25% to allow for the next five years’ expansion. For online designs (above 10kVA) future expansion can be achieved using parallel architectures.
An uninterruptible power supply should alleviate data centre managers’ headaches, not add to them and with a little prior planning and forethought, they can.
Data Centre managers have a lot on their plates. Pressured into squeezing more equipment into increasingly overcrowded spaces, energy-efficiency is also high on the agenda. Finance departments demand that costs are cut but that systems are kept up and running. In the face of a growing number of power cuts, it’s a wonder they can sleep at night!
Installing an uninterruptible power supply (UPS) need not add to the nightmare, as long as a few preliminary steps are taken.
What managers want from their UPS is: maximum uptime, system resilience and efficiency, low running costs and energy consumption, high density and a small footprint. How do they get that? This article tries to answer that question.
The Primary functions of a UPS. An uninterruptible power supply, primarily, provides a valuable source of back-up power in the event of a mains failure. It allows businesses to ride through short-duration interruptions; perform a controlled shutdown of non-critical systems and power loads whilst a back-up generator starts up.
A lesser-known function of UPS (but increasingly important one in data centre and server farm environments) is power conditioning. The voltage of raw mains energy can fluctuate wildly, causing power problems that are harmful to electrical equipment, particularly servers. UPS monitor and regulate mains power supply, keeping it to within acceptable levels and negating these harmful problems, thus saving on costly replacement or repair and data loss or corruption.
Maximising Uptime.Maximum uptime is achieved, firstly, by sizing loads and UPS correctly. This involves determining load power requirements, which can be confusing. On equipment labels and in user manuals, hardware manufacturers rate power usage unrealistically high. It is not uncommon to see loads running at only 50% of this rating. In addition, ratings given may be in Amps or Watts. Depending on the type of load, there can also be a significant difference between in-rush (start-up) and running power.
It is important, therefore, when sizing UPS to carefully consider: apparent power (VA), active power (Watts) and Power Factor (pf). The higher the input power factor, the more efficient the UPS will be. Today’s standard is upwards of 96%. Matching the right UPS topology and design to the specific application is also important, so too ensuring resilience and no single-point-of-failure. Additional runtime and/or some form of alternative power generation (a diesel generator, for example) and ensuring the whole power protection solution is consistently and properly maintained will add to the likelihood of 100% uptime.
Building-in System Resilience. System resilience (fault tolerance) comes largely from the UPS design and topology and how the system is configured. The primary components of resilience are the automatic static transfer switch that monitors the output waveform of the uninterruptible power supply’s inverter. Should it fail, due to a short-circuit, overload or fault condition, the switch transfers the load to mains supply without a break and automatically transfers it back once the fault has cleared.
Other levels of resilience come from whatever source of back-up power (batteries, diesel generator and/or a dual input power supply) is chosen.
Extended Runtime. When extending the runtime of an uninterruptible power supply, special consideration must be given to load category and type. Load category determines how critical it is to the continuity of business. Critical and essential loads need extended runtime and some form of redundancy. Non-essential loads (canteen facilities or printing services, for example) may need a solution to power emergency lighting (for health and safety whilst they are shutdown for the duration of the power cut) but no extended runtime. Runtime is load dependent: the bigger the load, the more power it will need.
Load ‘type’ refers to whether it is linear or non-linear, and how it draws its current from the mains power supply. It will be inductive, capacitive or resistive and this will have a bearing on UPS sizing.
Low running costs and energy consumption. Leading UPS manufacturers, such as Riello UPS, have made great strides in recent years towards providing the market with energy-efficient products. UPS energy-efficiency depends on: the energy they need to power them; the way they draw that energy; their affect on upstream supplies and the management of consumables (such as battery sets and switchgear). Modern UPS designs offer tangible improvements in all these areas, along with an ever-improving carbon footprints.
The latest zero-impact designs, such as Riello’s Multi Plus, demonstrate low input current distortion, reducing the need for dramatically oversized supply (generators and upstream switching). They give a near Unity power factor sinewave, leading to less wasted energy and are physically smaller, thus allowing more revenue-earning, high-density servers to be fitted onto floor space previously required for UPS and batteries.
Plan for the future – today! Data centres and server farms are fast-moving environments where growth and expansion are constant. When choosing a UPS solution for today, thinking about tomorrow (specifically expansion plans) is just as important. The most common approach is to ramp-up load sizing by 25% to allow for the next five years’ expansion. For online designs (above 10kVA) future expansion can be achieved using parallel architectures.
An uninterruptible power supply should alleviate data centre managers’ headaches, not add to them and with a little prior planning and forethought, they can.
