To invest or to upgrade… that’s the question

For most enterprises, it is difficult to decide whether they should invest in a new data center or should simply upgrade an existing one. By Manjari Juneja

For enterprises determining the future of their IT strategy, the choice of a new data center vs. upgrading the existing one depends upon the existing data center’s average power density per rack, power & cooling system reliability and its energy efficiency, and also the amount of headroom left. Once these facts are noted, they need to be correlated to the growth forecast for the IT application infrastructure of the enterprise.

Depending upon the circumstances, every enterprise should look at three avenues. Firstly, anything that it can do to extend the life of the existing set-up puts off its need to spend significant amounts of capital on building a new data center. Extending the life of the current data center is something that a company should be able to do relatively quickly and yet this is a process that would have an enormous financial impact on the business. Existing data center assets are paid for and depreciated, to some extent. So the first step should always be to try to squeeze out the maximum from the existing set-up and try to find ways to give it an additional lease of life.

Secondly, if the existing data center infrastructure is distributed globally, companies should look at rationalizing the current data center infrastructure and frame a strategy that will meet the needs of the company in terms of growth and resilience over the long haul and at the same time result in significant cost savings over the existing environment.

Thirdly, when customers have extended the life of their data centers to the limit, if they have rationalized their data center infrastructure on a global basis, they may still find themselves needing to build new data centers. If companies are going to build data centers today, they should design for the flexibility that they need at the lowest capital and operating cost that they can achieve.

Companies often misunderstand their own requirements and how to leverage the best from their existing IT infrastructure. Many organizations are underutilizing existing assets. As companies rethink their data center architectures to address these challenges, consolidation and relocation often provide valuable solutions. However, not all consolidation efforts deliver the same benefits. Comprehensive planning reduces complexity and brings significant service-level, cost and operational improvements during and after the consolidation.

A holistic approach is needed focusing on both the information required to run the business and the physical infrastructure.

Akash Saxena, Country Manager - Infrastructure Services, IBM India/South Asia, said, “Substantial improvements can be realized by improving the efficiency of the data center infrastructure by focusing on driving out inefficiency in the power conditioning and distribution systems as well as the cooling and lighting systems. Simple aspects like increasing the set point of air conditioners can also contribute in a big way. However, large improvements are usually possible only with an overhaul of the data center.”

S. Sriram, CEO, iValue InfoSolutions Pvt. Ltd., said, “Enterprises today have a third option of opting for hosted and managed third party data centers with SLAs. A data center’s life is typically four to six years and not many enterprises can manage to earn a return on investment. Compelling new trends such as server virtualization, green computing etc. can help an enterprise drastically reduce its OPEX. Therefore, if the need to have an on premise data center is compelling, it is a situation of constant mini upgrades and, once in a while, an overall redesign. Also with cloud computing being the next wave, dependence on a company’s own data center can reduce over time for many enterprises along with the benefits of avoiding CAPEX, lowering OPEX along with minimal overheads on the management part with third party options. When it comes to upgrade vs. setting up a new data center, it depends on how the existing data center is built and what server types are provisioned. When it comes to upgrading a data center, options such as reducing the footprint by considering server virtualization can have an impact. If designed in a modular way, we can add capacity, cooling, power, fire/smoke detection and emergency exits as per requirements. Any CAPEX investment which results in the reduction of OPEX along with a return on investment in less than 15 months should be prioritized. Drastic changes like JVs and acquisitions can impact this timetable adversely.”

Existing capacity

As per Gartner, the total data center capacity in India will increase from 1.337 million square feet in 2007 to 5.143 million square feet by 2012 at a CAGR of 31%. This is partly fueled by the fact that India is expected to be the data center hub for nearby markets such as the Middle East and South East Asia. There are even instances of European customers opting to host their data centers in India.

Nareshchandra Singh, Principal Research Analyst, Gartner, informed, “Data Center capacity is currently pegged at 3 million square feet and is expected to grow at a rate of 30% per year. Depending on the requirements, companies can go for a primary data center that is located at the company’s headquarters, and a secondary data center located outside the city that can be used for disaster recovery as well. While going in for an upgrade or replacing a data center, one needs to keep location, geo-political perspective, infrastructure cost, availability of power at an affordable price and the law & order situation of that place in mind.”

Approach to take

IT heads should begin with formal data center strategy development. It must be ascertained as to how long the present/planned data center facilities would satisfy the company’s IT needs. They must quantify uncertainty taking into account industry and proprietary IT trends and establish a baseline of what runs where, application growth model and decision model for IT and data center physical infrastructure.

Secondly, they have to design or upgrade for achieving high energy efficiency (the power usage effectiveness or PUE ratio) for power & cooling infrastructure. Designing for flexibility and deploying for the immediate future should be the approach, implying the need to grow in a modular fashion and to standardize each module.

It is recommended that the team always start with a clear documented statement of requirement for data center design. The data center user team, IT planning team and specialized firm for building the data center should form part of core steering team for these projects.

All design trade-offs should be done with data center TCO for its life cycle and the enterprise data center strategy in mind. Focus on data center operational cost and simulation of TCO during design trade-off decision making will always pay off.

Sanjay Lulla, Director, Technology Solutions - India and SAARC, EMC Corporation, said, “The goal is to feature efficiency and flexibility, make better use of existing and new infrastructure and resources, and to provide headroom to keep up with ongoing growth while minimizing the environmental burden. The top two points to keep in mind are to focus on re-architecting needs with a vision of the internal cloud and eventually private cloud as well as on information infrastructure elements rather than addressing individual symptoms.”

Some of the key considerations when transforming an existing data center or building new data centers are:

• An integrated virtual data center strategy that combines the integrated management of key operational processes and procedures, a flexible virtualized architecture and support for mission critical applications is essential.
• While moving from silos to services, take a step back to look at service-level requirements across the enterprise. First, think about the service requirements that your stakeholders are demanding. Then, compile these requirements into a catalog of IT services. Finally, map applications into service tiers.
• Protecting information assets from loss, theft, or misuse and ensuring that they are highly available and recoverable is increasingly important.
• Look at the maturity of your organization and data center operations by evaluating several areas: customer and service, organization and process, infrastructure and toolset, people and skills, and benefits realization. Then you can determine the priorities for improving these areas based on what will deliver the greatest benefits to the business.
• Today’s high-performance equipment has densely packed electronics in a compact footprint. This requires greater heat dissipation and higher cooling requirements. Therefore, it is important to pay close attention to the placement of densely populated racks in the data center. Misplacement of equipment results in higher power and cooling usage and higher energy costs. Organizations can take some simple steps to reduce cooling requirements and costs. These include creating hot aisles and cold aisles when configuring equipment placement.
• When considering a green data center initiative, begin with an inventory of all IT assets to assess and understand current power usage patterns. For increased assessment control and accuracy, consider logical division of the data center into more than one section, where power and cooling readings can be made in each section. Similarly, a data center asset division can be made by asset type such as servers, networks, or storage resources. An energy assessment should also clearly identify the energy and cost-savings over varying lengths of time.
• The key objective when designing an optimal data center cooling system is to create a clear path of air flow from the source of cooled air to the intake of equipment, and from the hot air exhaust of the equipment to the exhaust system’s return air duct.
• A regular health-check inspection ensures that the cooling demands in the data center are sufficiently met through the full potential output utilization of the CRAC system.
• Data center CRAC units must align with equipment placement. Once equipment aisle (hot aisle/cold aisle) configurations are determined, plan the placement of CRAC units inline with and on both sides of hot aisles.

Avoiding false starts

Addressing symptoms (for e.g. growing backup size) without considering the root cause which might be duplication of data is an example of a false start. There could be many more.

Gaurav Ahluwalia, Managing Director, R&M India, said, "70% of interruptions are caused by infrastructure and mishandling. One of the most common mistakes made by designers is failing to analyze the requirements in contrast with space availability and infrastructure layout. Often this process is neglected as designers tend to work with drawings first."

Some false starts that should be avoided are:

• Designing a data center with capacity deployment on day one for IT growth for the next 15-20 years.
• Designing with a capital budget focus rather than focus on total cost of ownership for the data center lifecycle.
• Start detail design activities without formal data center strategy mapped to the IT growth forecast.
• Lack of involvement of data center experts or specialized and experienced firms in the project design and execution stages.
• Lack of a clear and detailed statement of requirements as the starting point.
• No governance process to make the design trade-off decisions quickly in a fact based environment.
• The project being viewed as a real estate project with the technology and management systems having marginal involvement.
• Selecting an existing building as the site even before the data center statement of requirements is drafted.
• Power and cooling system designed for lower reliability level like 2 or 2+.

Srikant Chakrapani, Director, Enterprise Solutions, Hitachi Data Systems, said, “Today, the majority of data centers are configured in stove pipes. These stove pipes consist of applications which run on a server which is attached to storage. While servers may be clustered and storage may be connected in a SAN, these stove pipes do not share information between applications, do not move applications between servers, and do not move data between storage systems. This is not a dynamic data center. That is about to change with services oriented architectures, services oriented infrastructure, and services oriented storage. What is required to turn a stove pipe data center into a dynamic data center is a services orientation when it comes to applications, infrastructure, and storage. The dynamic data center of the future will be a services oriented data center based on SOA, SOI, and SOS. Virtualization is the foundation for any services approach by enabling the sharing and mobility of information, applications, and data across stove pipe infrastructures.”

Challenges involved One of the key challenges, especially for those creating their first enterprise data center, is when they take on the responsibility to manage the overall project without the necessary data center centric skills or experience. It is vital for them to not view a data center project as a real estate project. They should hire the services of IT Infrastructure specialists with the capability to deliver services starting from strategy & consulting to build & operate. A data center is a mission critical space and it needs design standardization and modular deployment to bring the best data center operation to a state of excellence.

Some key challenges are:

• Reliability: Data center downtime can cost between $50,000 and $6 million per hour. Designing solutions that offer redundant, fail-safe reliability is a must.
• Manageability: Disaster recovery, upgrades, relocation, modifications and the ability for future scale-ups are some of the key issues that should be kept in mind.
• Space Saving: Proper utilization of space is vital given that it is always scarce.

Ashis Guha, Country Manager - Brocade India, said, “A lack of holistic data center deployment strategy and the failure to consider the overall impact of virtualization on the rest of the data center network can cause more harm than good, effectively negating the cost, complexity, and flexibility benefits that the virtualization project was hoped to have achieved. By ignoring the impact that server virtualization has on the storage environment, the data center network, and users, organizations can experience management complexity, application degradation, inadequate business continuity policies, and, ultimately, increased costs. The data center also need to ensure that robust data protection and security practices are extended to virtual deployments. Often, organizations tend to backup and secure virtual machines as if they were physical servers, despite the increased traffic levels and mobility. At the same time, shared storage networks are required to support a growing number of applications due to server virtualization, putting additional pressure on an organization’s security strategy. This issue needs to be addressed by extending fabric services to virtual environments, ensuring a reliable and robust business strategy.”

Virtualized data centers

By employing virtualization in the data center infrastructure, organizations can take the first step toward achieving the flexibility and cost benefits of cloud computing. By accelerating virtualization deployments, an organization can boost its return on investment and improve the agility of its dynamic computing platform.

To realize the full benefits of virtualizing its data center infrastructure, an IT organization must have an integrated virtual data center strategy—one that integrates the management of key operational processes and procedures, a flexible virtualized architecture, and support for mission critical applications.

Virtualization in the data center focuses on increasing the utilization of IT assets, reducing variability of IT management and support, and infrastructure simplification. This requires implementing virtualization technologies to drive up utilization and drive down management and energy costs at every layer of the IT infrastructure stack (applications, OS, server, network and storage). For e.g., the objective of physical consolidation of servers in a data center is to reduce the number of hardware environments required, each with its own management systems, change control systems, maintenance schedules and unique environmental requirements. This will therefore reduce the number of instances where each application has to have its own dedicated server if security, governance and application constraints allow. There is also the need to reduce the number of servers performing the same operation be it a productivity or a line of business application. For e.g. it could help reduce the number of file and print servers in a company. Overall, virtualization in the data center helps in improving asset utilization and infrastructure capacity through server, storage, network and application virtualization services.

Ravichandra G.B., Director - Enterprise Business, APC by Schneider Electric, said, “Virtualization essentially allows one computer to do the job of multiple computers, by sharing the resources of a single computer across multiple environments. Virtualization is a software technology that is rapidly transforming the IT landscape and fundamentally changing the way that people compute. Virtual servers and virtual desktops let you host multiple operating systems and multiple applications locally and at remote locations, freeing you from physical and geographical limitations. In addition to energy savings and lower capital expenses due to more efficient use of hardware resources, customers see higher availability of resources, increased security, and improved disaster recovery processes when virtual infrastructure is built. However this has to be coupled with appropriate and right sized power and cooling for optimum results and efficient computing or else the data center efficiencies can get affected.”

The value of a virtualized infrastructure lies in the increased flexibility created, by having pools of system resources upon which to draw; in the improved access to information afforded by shared infrastructure; and the lower total cost of ownership that comes with decreased management costs, increased asset utilization, and the ability to link infrastructure performance to specific business goals.

manjari.juneja@expressindia.com