Green data center: Economy Vs Environment

Green data center: Economy Vs Environment

Green data centers don't just save energy; they reduce the need for expensive infrastructure upgrades to deal with increased power and cooling demand

Automation and networking of state and central departments is the first step for better governance and India's National e-Governance Program (NeGP) is a paradigm shift in service provisioning-from government-centric to citizen centric.

The plan is to start treating citizens as government customers and empowering them to demand convenient, cost effective and transparent services.

For this NeGP has identified three core infrastructure components - State Wide Area Network (SWAN), Common Service Centre (CSC) and State Data Centre (SDC).

Apart from this it comprises of several mission mode projects (MMP) to be implemented by line ministries and departments at central government or state governments, as well as integrated projects spanning multiple ministries, departments and agencies.

In last three months, 10 state governments-Tripura, Maharashtra, Haryana, Orissa, Meghalaya, Puducherry, Gujarat, Sikkim, Nagaland and Rajasthan have floated tenders for design, site preparation, supply, installation, commissioning and maintenance and operations of State Data Centers (SDCs) for a period of five years.

This number is bound to increase as many states are in advanced stage of data center Request for Planning (RFP). All these deployments will help in giving big boost to e-governance in these states.

SDC is envisioned as a shared, reliable and secure infrastructure for hosting and managing e-Governance applications of state and its different departments. It is a robust infrastructure and enables government to deliver services quickly to all its stakeholders.

It provides functionalities such as central repository of the state, secure data storage, online delivery of services, citizen information and services portal, state Intranet portal, disaster recovery, remote management and service integration.

With these shared service centres, implemented and managed by a competent agency, individual departments can now focus more on service delivery rather than on issues with regards to infrastructure.

Till yesterday, state governments had multiple data centres specific to different departments. But this scenario is changing as consolidated data centres are the need of the hour as it help in judicious use of hardware and software, as also power, security and maintenance requirements.

However, shortage of power has emerged as a key problem in deployment of an efficient data center.

Also, with increasing recognition that man-made greenhouse gas emissions are a major contributing factor to global warming, enterprises, governments, and society at large now have an important new agenda: tackling environmental issues and adopting environmentally sound practices.

Technology service providers like Cisco are trying hard to make its IT products, applications, services, and practices more and more environment conducive which have now become an economic and an environmental imperative, as well as our social responsibility.

As there have been acute power shortages in all the states, there is a crying need to use power judiciously across different applications.

Reducing power consumption in data centres is, therefore, a priority for state governments and they can opt for green data centres as they consume less power and provide significant environmental benefits.

The long-term goal of the green data center operation is to achieve carbon neutrality and also provide energy efficiency.

IT managers can design power and cooling efficiencies in their data centers through the strategic application of Cisco and its partner technologies.

Blade Vs Rack Unit
In recent years, lot of attention is focused on blade form factors in server platforms. Blade form factors per work unit performed consume power and cooling more efficiently than stand alone rack unit servers.

Data center efficiency is in part a cumulative calculation of how disparate IT infrastructure interoperates with supporting facilities, on top of this blade capacity and density requirements must be aligned with facility design to achieve maximum efficiency.

Power Densities and Facilities Design
Any data center manager can attest that a data center is almost never deployed uniformly after the first year.

Adhering to a watt density per rack helps give the data center manager more control over the standard operating environment (SOE) per rack.

From a planning and deployment standpoint, this variable is a much more manageable building block that can be used in a zone-based approach.

The practice of data center zoning is particularly important in a virtualised environment. For example, a data center may have 100 servers that support three very different applications.

Each of these applications has different criticality requirements, but all may be running in any micro environment within the data center at any given time.

Any of these microenvironments may be peaking over the specified temperature thresholds, regularly causing a threat to the availability of servers.

Cooling Unit
The largest consumer of power and the most inefficient system in the data center is computer room air conditioning (CRAC).

Almost half the power in data center goes to cooling, mainly because of the burden factor of air conditioning, which is the ratio between production power and cooling power.

This ratio typically ranges from 1.8 to 2.5, depending on many factors. For example, if a server requires 100 watts of production power, a CRAC unit would require between 180 and 250 watts of power to cool the server.

The factors that contribute to the burden factor of a CRAC unit are mainly the operational efficiency and air distribution design of the unit or how efficiently the air can reach the server it is cooling.

Examining air flow in the data center typically uncovers many simple fixes that can increase the efficiency of the entire cooling system. Cisco and its partners help customers implement many best practices.

In some cases, these fixes are as simple as installing blanking panels in open rack space, rearranging perforated tiles, and installing hot-air and cold-air partitioning.

In other cases, they may be more complex retrofit projects such as migration of cabling out of the subfloor plenum to eliminate constraints to air flow.

The best time to design efficiency into an air-distribution strategy is when building a new data center that can accommodate densities in line with current IT infrastructure projections.

Rack Environment
Rack is the primary building block for power and cooling planning and watts can be adopted as the unit of measurement for both power and cooling of a given rack, simplifying the higher-level planning for rack deployment.

For air flow within the rack, a cfm (1 cfm equals about 28.31 liters per minute) measure should be employed because air distribution is as critical a factor as air supply.

Using a watts-per-rack measurement standard, the data center manager can plan more effectively.

A watt per rack standard lets a manager set standards and policies for incremental additions of IT infrastructure and better plan how to scale the facilities infrastructure to meet IT demand.

IP Management
The facilities infrastructure, including uninterruptible power supplies (UPSs), computer room air handlers (CRAHs), temperature and humidity monitors, fire suppression, physical security, power distribution units (PDUs), and branch circuits, are still not monitored in an IP-based network.

This poses significant challenges in planning incremental additions but significant advances has allowed IT managers to monitor, measure, and manage physical layer through an IP network.

Implementing a network-based monitoring strategy allows the data center manager to measure progress toward goals to improve power efficiency.

One can identify many efficiency gains. For example, providing cooling on demand is always more efficient than overcooling.

Proactive network monitoring delivers the data needed to make more informed decisions about how to achieve efficiencies in the system.

Data center efficiency losses are roughly 80 per cent process and 20 per cent infrastructure losses, IP-based network management can have the greatest effect on data center efficiency at the operations level.

Network Architecture
The data center efficiency must be quantified at systems level and needs to be aggregated for the entire site.

When considering network architecture, the planner should not confine the scope to switches, routers, load balancers, and firewalls and should extend it to gather as many relevant data sources as possible to monitor power efficiency.

These sources include monitoring of power, temperature, and humidity; UPS, branch circuits, PDUs, specialised sensors and others that are located either inside or close to the data center.

In addition to using the network, efficiency opportunities in network design to support different services are significant.

If application delivery, security, access and identity, storage, and computing are considered as services, the network can have a profound effect on the cumulative efficiency of an operation.

Network Effect
The network effect refers to the collapsing of services or portions thereof provided by standalone components into a more ubiquitous network-based platform.

Numerous studies have indicated that server and storage utilisation rates range from 20 to 40 per cent.

Of course this level of inefficiency in asset utilisation has a corollary effect on power usage in a given operation. Taking advantage of technology, utilisation should be increased.

Need of the Hour
Data centers or SDCs for that matter is slowly but gradually becoming the hub of IT-enabled service delivery system for the government agencies.

In today's context neither private bodies nor government agencies can afford to use its resources liberally.

A judicious utilisation of its resources not withstanding service delivery efficiency is the need of the hour.

By Pravin Prashant, Editor,