White Paper. Control room convergence: Merging industrial monitoring and control systems with data center operations

Transcription

1 White Paper Control room convergence: Merging industrial monitoring and control systems with data center operations

2 Data centers as an industry Until recently, data centers were not regarded as an industry unto themselves. Even the term is relatively new the same facilities were known simply as computer rooms as late as the 1990s. The rapid growth in the size, number and sophistication of data centers in the past 20 years is staggering, and there is now a robust industry dedicated to serving this market. Growth in data centers is being driven by rising demand for computing and storage capacity that is in turn the result of several parallel trends. The proliferation of mobile applications, for example, has created an entirely new category of software in just the past few years. These programs themselves are not particularly sophisticated, but when you multiply their rapidly expanding numbers by the ever-growing number of smart phone and tablet users, the result has very real and immediate implications for the data center industry. Projected increase Number of data center facilities: 7% Number of server racks: 15% Energy consumption: 19% Source: Datacenter Dynamics Global Industry Census 2011 Looking at the industry itself, much attention has been given to top-tier players such as Google, Apple and Facebook, and understandably so. These companies are investing hundreds of millions of dollars in purpose-built, highly engineered facilities that represent the state of the art in data center design. But for every one of these marquee installations there are hundreds, probably thousands, of more average data centers that are facing some difficult challenges, such as the need to reduce operating costs, manage an ever-expanding inventory of equipment and monitor operations across multiple facilities. Data centers are not merely computer rooms on a larger scale. They are highly specialized industrial facilities, no different from chemical plants or oil refineries in terms of their particular mission-critical needs. Still, for the most part, the tools being used to manage these facilities are not in line with the demands being placed on them. We are now at a pivotal moment in the history of the data center industry. Vital work is being done now to address the sector s needs, but to understand how this will lay the foundation for the industry going forward, it s helpful to examine its recent past. Evolution of the data center In the early days of the internet prior to 2000, data centers had little or no unified monitoring capability. In many ways they were simply large computer rooms cold, windowless spaces stocked with rows of humming servers. Meters were placed on various pieces of equipment, but readings had to be taken by hand. There was no way to know in real time how much energy was being used by the facility, much less by a given piece of equipment. It was equally difficult to know how much a given server was actually being utilized. There was also no pressing need to know energy was cheap and capacity was not an issue. As IT took on greater importance in the business world and the web developed into the most powerful commercial, social and technological engine we have today, businesses scrambled to scale up their data centers to handle the additional traffic. Energy consumption is one area that provides a microcosm of the rapid evolution data centers have undergone. Just five years ago, typical servers drew around 2.5kW per rack, but today servers consume between 8 kw and 30 kw per rack. They still occupy the same amount of space, which means that cooling has become an even greater challenge. The combination of higher energy density and the associated increase in cooling load has pushed data center energy use to unprecedented heights. The industry has responded to advances in server technology with advances in data center operations, and many best practices have emerged. Many data centers 2 Data Centers ABB White Paper

3 have implemented hot aisle/cold aisle schemes, for example, which more effectively manage air flow to avoid hot and cold air mixing. Some facilities in colder climates have also utilized free air cooling to take advantage of cooler outside air. However, these are passive solutions. When it comes to monitoring and control, the systems in place today have not kept pace with their counterparts on the hardware side. Point solutions are available, for example to manage server virtualization or monitor energy use on a facility-wide basis, but importantly, these systems are not integrated, thus adding complexity and gaps in information that are susceptible to human error. Projected 2012 Data Center Investment by Country/Region Country/ region Investment (BUSD) USA UK 3.35 China 3.10 Germany 2.60 Australia 2.45 Brazil 2.15 France 1.95 Italy 1.95 Canada 1.90 SE Asia 1.90 Mexico 1.80 Turkey 1.80 Russia 1.65 Argentina 1.60 Benelux 1.55 India 1.50 Middle East 1.50 Spain 1.35 Colombia 1.20 Nordics.95 Source: Datacenter Dynamics Global Industry Census In other areas, data centers still lag other industries by a wide margin. Asset management, for example, while having many definitions is generally understood to mean the systematic monitoring of equipment status and performance to better manage maintenance and optimize O&M budget across the entire equipment fleet. In the data center world, asset management has historically meant simply keeping track of the physical location of equipment in the building. Such is the reality of an industry experiencing exponential growth. Today, the data center industry is going through a kind of adolescence. ABB estimates that around 5 to 10 percent of data centers can be considered advanced, having monitoring and control systems in place for server operations, energy consumption and environmental control. Another 20 to 40 percent are in the average range with some monitoring and control capability most likely around server operations and the remaining 40 to 65 percent have little to no monitoring and control capability. Part of the reason behind this disparity lies in how data centers are built. Historically, they were treated as just another commercial office space, albeit with some unusual requirements for electricity. New facilities are more customized but a large number of legacy data centers remain. One by-product of this is the fact that many, if not most, older data centers rely on conventional building management systems to control temperature and humidity. Consequently, while the state-of-the-art in data centers is advancing, there is a large need for retrofit. ABB White paper Data Centers 3

4 What is DCIM? Enter data center infrastructure management (DCIM) In light of the above, it has not gone unnoticed that there is a need for a unified monitoring and control system for data centers. Several vendors including ABB are now bringing solutions to market that can provide ready access to normalized data from across the enterprise via an intuitive user interface. This field of technology development has been dubbed data center infrastructure management. From a strategic standpoint, DCIM is about aligning the data center with business goals and realizing greater value from it through: - energy and performance information, - operational decisions supporting business initiatives, and - delivery of new apps and web services in the most reliable and efficient way pos- _sible. Asset management in the data center world means simply keeping track of where equipment is in the building. The tools currently available for server operations are already quite sophisticated, but as we ve noted, those for managing environmental parameters and energy use in data centers are less so. The monitoring and control technologies currently available produce multiple sets of data that must be managed individually. This has created a tremendous opportunity for automation. Operating costs are now coming to the fore as a major concern for data center operators, and while operating efficiency occupies a greater share of mind (at least in North America) than energy efficiency, in fact the two are closely related. What the industry is looking for now is a system with a single user interface to reduce the complexity of burgeoning point solution interfaces, and the ability to manage server operations as well as cooling, environmental controls and energy use at various levels within the facility. The good news is that other process industries have already tackled many of these same issues, so the data center industry is in a position to leverage those mission-critical systems in building its own. Decathlon Utilization Analysis Dashboard 4 Data Centers ABB White Paper

5 So, what does a DCIM system look like? Several characteristics are essential. Above all, DCIM systems must be highly reliable. They have redundancy at both the component and system levels, and they undergo rigorous integrated testing at both of those levels. Uptime is the lifeblood of the data center industry, so availability is vital. In terms of features, DCIM systems offer asset management capabilities that go well beyond simply keeping track of servers. Condition-based maintenance and diagnostic tools (e.g., to identify servers running in a loop) will bring asset management in the industry into line with other equipment-intensive businesses. As more operators are faced with managing multiple data centers, computer rooms and server closets, multi-site visibility is another must-have for DCIM. This will take on even greater importance in the future, too, as we will discuss later. DCIM is ultimately about providing the visibility, decision support and control technologies to better manage data center operations across the enterprise, worldwide, through a unified view and single data source. That, in turn, allows the organization to deliver services faster, in the most reliable and efficient way possible. DCIM systems promise to deliver actionable information to data center operators so they can maximize capacity, optimize their operations, and reduce cost and risk. DCIM enables data center operators to manage operations across the enterprise, worldwide, through a unified view and single data source. Following are a few examples of different areas where DCIM offers a significant improvement over the status quo. For example, Decathlon, the data center infrastructure management solution by ABB, provides the following insight: Temperature/humidity. Sensors onboard each server provide a fine grain view of temperature, and others provide humidity readings from throughout the facility; trouble areas can be identified and addressed specifically. Environmental control. Monitoring and centralized control over fan speed and chillers allows operators to optimize performance of individual units and perform trending analysis. Decathlon Rack Information View ABB White paper Data Centers 5

6 CPU utilization. Data collected directly from servers in real time provides an accurate account of how the data center is using its computing resources. Power management. Metering, even at the device level, can be managed centrally. This enables automated decisions based on a variety of inputs (e.g., weather forecasts, utility contracts, server loading patterns, etc.) to optimize energy use. On-site generation/storage. Rather than simply an emergency backup, on-site generation and energy storage can be used in conjunction with grid power to optimize costs. Going forward, DCIM will enable data centers to take advantage of demand response programs and ultimately energy arbitrage across multiple locations. Maintenance. DCIM enables a shift from conventional time-based maintenance schemes to condition-based maintenance, enabling a prognostic vs. diagnostic approach. DCIM systems can also be configured to diagnose potential problems from sub-critical alarms. Remote access. DCIM allow technical experts to serve multiple facilities remotely, making them more cost-effective (e.g., to deal with alarms). The case for DCIM As we ve discussed, the need for visibility, decision support and control technologies for rapid response in data centers is clear and DCIM offers a solution to the patchwork of point solutions that most facilities rely on today. DCIM is still evolving, but a few overarching concepts have emerged that are likely to guide the development of these systems going forward. The essence of the DCIM business case is improving visibility and control of data center operations. Visibility and better control. The essence of the DCIM business case is improving visibility and control of data center operations. It starts with establishing a realistic baseline of data center energy use through real-time, comprehensive monitoring, so operators know when, where, and at what rate energy is consumed. Armed with that information, they can then take tactical measures to optimize resources and forecast energy requirements. It is also worth noting that a DCIM system should be user friendly to facilitate visibility and better control. Resource Consolidation. DCIM systems leverage increased visibility along with powerful analytics to consolidate resources and eliminate waste, thus maximizing existing capacity. This leads to hardware savings (e.g., fewer servers), data center savings (e.g., power, cooling and space), and reduced environmental impact (e.g., lower carbon emissions). Performance Optimization. DCIM uncovers the true state of data center operations with better visibility and control to improve availability; maximize capacity of power, cooling and space; streamline operations and reduce risk across the enterprise. Examples include condition monitoring to drive maintenance, maximizing efficiency by analyzing loading and virtualization of servers, and identifying idle servers that can be removed. Facility and IT Automation. DCIM systems automate facility and IT operations for more rapid service response through dynamic IT (e.g., load shifting) and automated workflow processes. They also monitor and control temperature and humidity pro-actively, and track other parameters such as vibration in HVAC units that can be used to identify potential equipment failures before they occur. In many ways, the value proposition of DCIM is very similar to that of automation in 6 Data Centers ABB White Paper

7 The future of data centers general, which has guided process industries for more than fifty years. The ultimate objective is to use information to optimize system-wide performance while minimizing cost and risk. Decathlon Floor Layout View What s next? As noted earlier, the data center industry is a highly dynamic one that is presently experiencing certain growing pains as it strives to meet ever higher expectations. For the vast majority of facilities that do not enjoy the advantages of cutting edge systems and design, a step-wise approach to improving data center operations is advisable. The first order of business should be a site evaluation. This will reveal where the most immediate gains can be made (e.g., in energy savings). It will also facilitate the eventual introduction of a DCIM system if operators know what they want to do with it. Retrofits should begin with simple initiatives with quick payback (i.e., less than one year). These might include increasing room temperature or installing variable speed fans in the cooling system. Going forward, data center operators will need to view their facilities in a more holistic way, and DCIM will play a vital role in that process. In fact, DCIM systems are already a competitive advantage among the technically advanced top-tier operators, and as they proliferate across the industry they will offer the same to data centers of all sizes. One example of where DCIM is headed lies in the potential for data centers to engage the power industry by leveraging their unique ability to shift workload from one location to another. The larger and more numerous the facilities at a data center operator s disposal, the greater the potential to exploit local differences in energy prices. This is something that old economy businesses can only dream of instantly shifting production from one plant to another but data centers are the epitome of new economy business. The application and further refinement of DCIM systems will play a central role in allowing this industry to realize its full potential. ABB White paper Data Centers 7

8 Contact us ABB Inc. Eric Olson Data Center Product Manager 125 East County Line Road Warminster, PA USA Phone: HELP AKK105408A9488 Document ID Marina Thiry Director, Strategic Marketing - Data Centers ABB, Inc. 730 University Drive Menlo Park, CA USA Phone: +1 (650) marina.thiry@us.abb.com