© University of Cancun, Mexico1 Chapter 8: Green computing and Communication Architecture 1 Tarik Guelzim, 1 Mohammad S. Obaidat, Fellow of IEEE 1 Monmouth

© University of Cancun, Mexico 1

Chapter 8: Green computing and Communication

Architecture

1Tarik Guelzim, 1Mohammad S. Obaidat, Fellow of IEEE

1Monmouth University

HANDBOOK ON GREEN INFORMATION AND COMMUNICATION SYSTEMS


Energy Consumption big issues

Enterprises, governments and institutions at large have a new important agenda: tackle the most challenging environmental issues Adopt environmentally sound practices.

Computers and IT infrastructure consume significant amounts of electricity, placing a big burden on power grids


Exponential growth of content

In order to cope up with the increasing demand for content, every year close to 150,000 new cellular towers are deployed across the world to serve around 400 million new subscribers worldwide.

Meeting this demand is becoming a burden on energy consumption if we take into consideration the energy cost which is skyrocketing in the middle of ever changing geopolitics and economics around the world.


Evolving broadband internet

it is no longer an option to focus only on the technological development of broadband internet to solve the capacity problem

it is no longer an option to hold on to the current networking communication protocols that have reached a road block with regards to their initial design and features


Green computing: Hype vs. Standard

To this day, there is more hype than real science behind the term green computing and communications which does not have a clear scientific definition other than its commercial and marketing connotation.

We need a standard to adopt, which considers efficiency in communication networks equipment along with the use of alternative sources of energy


Green computing and governmental effort

the study of green communications requires investigation in many areas such as radio frequency (RF) hardware, Media Access Control (MAC) protocols, general networking, material design and integration of renewable energy

metrics are key elements to properly measure and provide indicators to allow making better judgments and strategically plan the growth of the different communication networks infrastructure.


Green computing and governmental effort: Energy Star 4.0

The program energy star has provided administrative guidance in reducing energy and power consumption in devices, appliances both civil and military.

Energy Star is today one of the first technologies that were built in and used inherently in data centers. This because virtually every computer device today has this technology incorporated within and natively.


Green computing and governmental effort: Energy Star 4.0

The new energy star 4.0 standard regulates energy performance of external and internal power supplies.

It also gives power consumption specifications for different hardware states: idle, sleep and standby modes.

Computers meeting these requirements will save energy in all modes of operation.


Green computing and governmental effort: The TIP program

The TIP managed green communication program allows a new solution that addresses the inherent limitations of the vertically separated design of OSI layers.

TIP puts the emphasis on combining layer 1 and layer 2 (physical and MAC) to promote factoring research effort in one area


Green computing and governmental effort: The TIP program

It also promotes using scientifically rigorous definitions and metrics to allow agencies quantify and qualify green communication infrastructure.

More of a framework than a specific technology, TIP managed programs are being experimented with in fields such as transportation, medical systems and finance


Green computing and governmental effort: Electronics energy efficiency effort

The electronics industry has put a lot of effort on improving components efficiency because they are the basic building blocks of any system.

the Silicon Integration Initiative (S2I) and the Power Forward Initiative (PFI) are putting conventions for low power design by standardizing computer hardware interfaces.

Initial forecasts have shown promising that it is possible to reduce carbon emission by 50% over a decade


Green computing and governmental effort: Epeat program

The Epeat program, is a comprehensive environmental rating that helps greener computer and electronic equipment.

Epeat evaluates electronic products on 23 required criteria and 28 optional ones



Epeat requirements are grouped into eight categories: Reducing and eliminating environmentally

sensitive materials Selecting materials Designing for the products end of life (recycling) Product longevity Energy conservation End of life management Corporate performance Packaging



Epeat identifies its registered products as bronze, silver or gold. Bronze products meet all 23 required criteria

silver meets all 23 criteria in addition to at least 14 optional criteria

Gold class which meet all 23 required criteria in addition to at least 21 optional ones.

Optional criteria are freely selective by manufacturers


ENERGY BASED MONITORING ARCHITECTURE FOR GREEN COMPUTING :

Study of power Consumption in traditional PC architecture

The following graph depicts the energy consumption by different parts of traditional computer architecture under normal usage conditions



Energy Monitoring

More than ever, enterprise architects and deciders need tools such as indicators and dashboards to allow them successfully forecast the energy consumption of their future system (SCADA system, data center … etc).

Based on these forecasts, existing systems can be optimized.



Energy Monitoring

Tools / Technique Usage

Circuit meters Used in data centers especially in a “rack

group” setup. Circuit meters allow

measuring precisely the energy

consumption of a group of server racks.

Power strip meters Also used in data centers. Power strip

meters allow a group of interconnected

systems.

Plug meter The plug meter allows monitoring one

physical system.

External thermal sensor

meter

The thermal sensor meter allows a more

grained sensing capability at a floor level.

Internal server thermal

meter

Used in inside a physical system.



Extensible computing Architecture

Any green computing architecture today must take into consideration the environmental impact it incurs.

IT architect shall aggregate the communication models with the different metering interfaces from the ground up to allow a better synergy and effectiveness in terms extensibility in the long run.

There is a lack of standards and the segmentation of the different vendors about the very basic elements of computer architecture design




The following reusable architectural blocks have been identified to create extensible green computing architecture systems:

Issue Description

Proprietary energy APIs Most vendors have developed their own APIs to

interface and communicate with metering devices.

Especially in large IT structures, we can find more

than one interfacing API each corresponding to a

different system. This fragmentation phenomenon is

similar to having to use many cell phones and using

different battery charger for each one of them.

Inevitably, this incurs and multiplies additional

maintenance cost for the different architectures.




Issue Description

Environmental

communication bus

Because there is a heterogeneous metering

environment, system architects attempts to design

architectures that comprise a common

communication bus which acts as a façade when

communicating with the different subsystems.

Consolidating the communication bus makes it

possible to create better monitoring and reporting

tools.




Issue Description

Configuration management Configuration is an important aspect of any system. In large

systems, configuration changes can be costly (eg: number of

backups, number of allowed concurrent sessions, disk quota,

redundancy …) configuration can add work load on the

system. Configuration management should allow correlating

energy consumption to configured system profiles and also

allow adapting the system dynamically based on its GHG

measurement.



Best practices for sustainable green computing system design

The following are few best practices when designing a green computing architecture:

Understanding business objectives Understanding power consumption and

impact Developing a monitoring strategy Build environment sustainability into

configuration management



Incorporating Alternative energy in the basic BTS architecture

Several cellular operators are experimenting with the use of alternative energy such as wind and solar in order to operate cellular base stations

Currently, there are still many base stations in African countries that run on diesel fuel.

the Mobile Telecommunication Limited (MTC) and Motorola have initiated a trial to study the feasibility of using solar panels to fuel cellular base stations.

Results have suggested that approximately 659 kg CO2 could be saved by eliminating the diesel generator


Green Communication protocols and models: Cognitive radio

Recent advances and breakthroughs in the area of cognitive radio have significant potentials toward green communication.

Cognitive radio, as its name inspires, relies fundamentally on sensing the surrounding environment and readapting its behavior on the different elements collected

The integration of contextual “awareness” in the wireless spectrum opens a wide range of power control possibilities


Green Communication protocols and models: Cognitive radio

This technology does not however come without drawbacks, in reality, in today’s setups;

Using less power at the cell level must be compensated by more power consumption at the device level resulting in a zero sum gain

It is clear that an interdisciplinary method in the telecom field between cell tower and cell phone handheld manufacturers must be put into place to push forward an inter-layer approach.


Green Communication protocols and models: First order radio model (FORM)

The first order radio model has been developed in an attempt to solve energy inefficiency issues with sensor networks

Under normal conditions, FORM diminishes energy dissipation between the transmitter and the receiver nodes.

Energy dissipation per bit in nano-joules• 50 nJ/bit50 nJ/bit

Transmitting a message of k-bits a distance d radio expends

The receiving end radio expands as such


Green Communication protocols and models:Direct communication protocol (DCP)

This protocol is used for routing in combination with low energy dissipation model such as FORM. Using DCP.

Each sensor sends data directly to the base station.

Only the base station receives messages


Green Communication protocols and models:Minimum transmission energy protocol (MTE)

Nodes use intermediate nodes for routing to send data to the base station. The intermediate nodes path is chosen as such that it minimizes the energy amplifier equation (using FORM model).

This protocol has few drawbacks: if relied on minimizing the amplifier equation, all

nodes close to the base station will be chosen as intermediate nodes since they do not require a large amount of signal amplification to send data to the base station.


Green Communication protocols and models:Clustering

The clustering technique tries to marry the benefits of both DCP and MTP.

Under clustering, nodes are organized into clusters with a “delegate” base station.

These base stations then transmit consolidated data to the main base. This technique works best if the base station is not an energy constrained node


Green Communication protocols and models:Low Energy Adaptive Clustering Hierarchy

(LEACH)

LEACH improves on the previously described Clustering algorithms by distributing cluster information and by adding local processing to reduce global communication.

In addition, LEACH adds random rotation of the cluster heads to maximize and extend the system life time.

This protocol is broken into different phases: setup and steady state phase.


Green Communication protocols and models:Wireless Network Distributed Computing (WNDC)

The current traditional wireless model empowers a “selfish” node behavior that is solely based on: performance quality of service capacity in the surrounding areas that are covered

by its cell.

The requirements of each base station must rather be coordinated within the overall network


Green Communication protocols and models:Wireless Network Distributed Computing (WNDC)

The ability to apply such concepts in the wireless domain remains a challenge given the disruptive characteristics of the wireless channel.

Distributed Design tradeoffs involve the following: A cross system interaction between the

application layer and the communication layer. Simplifying the interaction complexity in the

underlying networking, radio and physical layers.


Green Communication protocols and models:System Power Consumption (SPC)

System Power Consumption is yet another communication model that aims at lowering energy consumption.

Recent research has proposed to limit the energy consumption by adapting modulation techniques, coding and radiated power.

This research works on the same energy optimization track as the PA model described in a previous section.


Green Communication protocols and models: Network Topology and Operation

One of the leading telecom companies, Huawai, is leading the way to provide energy efficient base transceiver station (BTS) by including PA improvement along with operational strategies

At the core of their technology is the Doherty-enhance technology that is combined with a custom power amplification chip to elevate amplifier efficiency to 45% instead of the current 33%.


Green Communication protocols and models: Dynamic Spectrum Access (DSA)

Dynamic Spectrum Access (DSA) provides the sensing and dynamic reconfiguration capabilities that leverage the use of spectrum white holes; this allows supporting opportunistic transmission without requiring extra spectrum bandwidth.

It offered solutions to many issues: Offering real time processing using low clock

frequency operation. Reducing RF accessing needs since this is a power

consuming factor. Processing multiple instructions with each clock cycle.


DESIGNING GREEN COMPUTING MODELS: Green processor design

Green computer design aims at reducing the environmental impact of computing by adopting new technologies and using new techniques and materials while balancing performance with the economic viability of the green solution

Some research suggests that 50% energy consumption gain can be achieved by only decreasing the chip frequency by 15%. Other initiatives concentrate on dividing the cache into segments and only enabling i.e. powering them when needed.


DESIGNING GREEN COMPUTING MODELS: Greening data centers

The continued rise of the internet and web applications is driving the rapid growth of data centers.

We can improve data centers efficiency by: Using new energy –efficient equipment, Improving air flow management to reduce cooling

requirements, Investing in energy management software Adopting environmentally friendly designs for

data centers


DESIGNING GREEN COMPUTING MODELS: Greening data centers

Some of currently deployed measures in the industry are: Energy conservation: Companies like HP, IBM, SprayCool a,d

Cooligy are developing technologies such as liquid cooling and nano fluid cooling into servers and racks. Others use green energy sources such as hydrogen fuel cells as alternative green power sources.

Virtualization: this is a key strategy to reduce data center power consumption. With virtualization, one server can host multiple virtual servers that use the computing resources efficiently because computing resources (CPU clocks cycles) are often not used at their maximum capacity.


DESIGNING GREEN COMPUTING MODELS: Green super computing

Data center supercomputers provide an unparallel computational horse power for solving scientific and engineering problems

This horse power comes at the expense of power consumption not only to run the super computer but also to cool it down.

supercomputing focus on performance and “occasionally” cost/performance where performance is defined as speed


DESIGNING GREEN COMPUTING MODELS: Green super computing

Green destiny is the first major instantiation of the supercomputing in small spaces project.

Green destiny has put into perspective two types of super computer architectures: A low-power supercomputer that balances

performance and power at system integration time. A power-aware supercomputer that adapts power to

performance needs when the system is running. Both of these approaches aim at reducing power

consumption and improving energy efficiency.


DESIGNING GREEN COMPUTING MODELS: Web services based SaaS and the Cloud

Software as a Service (SaaS) is a buzzword in the realm of cloud computing.

The basic idea is that we can reduce the number of applications deployed in data center by consolidating similar applications by SaaS providers

There are many major players in the market today such as salesforce.com and amazon.com who offer SaaS services at a low cost.


DESIGNING GREEN COMPUTING MODELS: Web services based SaaS and the Cloud

The major keys benefits of Web Service based SaaS can be summarized as follows: Cost (money) effectiveness. Time effectiveness. Focus on business needs rather than the

construction and maintenance of IT infrastructure. Gain immediate access to state of the art

innovations rather than maintaining legacy infrastructure.

Gaining instant benchmarking data based on community feedback of SaaS.

Save energy by reducing storage, computational and processing power at the servers


DESIGNING GREEN COMPUTING MODELS: Green Operating systems (OS)

There are many ways to optimize a client environment to save energy and reduce environmental impact.

Recent operating systems have made a clear design cut between the different layers that make up an –general purpose- operating system



There are many ways to optimize a client environment to save energy and reduce environmental impact.

Recent operating systems have made a clear design cut between the different layers that make up an –general purpose- operating system



Each layer has its own services and provides services to the above layer.

Each layer has to work in coordination with other layers.

This layering model, allows dividing the “green” requirement into three scopes, each one is handled by the contributor of the subpart in the final OS.


Conclusions

In this chapter we have summarized major research and technical orientations in the field of green communications and networks.

These orientations focus on: Energy efficient protocols Energy efficient signal processing methods Energy efficient Adaptive algorithms Energy dissipation in computation Energy efficient network architectures Alternative energy sources


Conclusions

Given the multidisciplinary nature of green communications that ranges from material engineering to system architecture, breakthroughs in this field are arriving at a fast pace.

Nevertheless, implementing sustainable practices and strategies in businesses, products and services is also as important as their engineering counterparts in order to achieve this milestone.


Conclusions

Ignoring power consumption as a design constraint of IT systems will result in supercomputing systems with very high operational cost and small reliability.

This is to say that future petaflops machine will require 75 to 100 MW to power up and later cool down which can be estimated in several millions of US dollars in a span of a year.


Conclusions

Green IT is a hot topic today and will continue to be an important issue for several years to come. To foster green IT, we need to understand the key environmental impacts arising from IT as well as the major issues that need to be addressed.

IT infrastructure, products and services, operations and applications must turn into “green” sustainable forms.


Conclusions

The challenges of green IT are immense. Nevertheless, recent development indicate that the IT industry is aware and has the “will” to tackle these issues given the enterprise business orientations are not impacted by such measures, at least not at the same instance in time.


Thanks for your attention!

Documents

© University of Cancun, Mexico1 Chapter 8: Green computing and Communication Architecture 1 Tarik Guelzim, 1 Mohammad S. Obaidat, Fellow of IEEE 1 Monmouth