Liv Stewart and Samuel A. Recine

The broad selection of IT components available to consumers has continued to grow by leaps and bounds in recent years. With respect to graphics processing hardware and software, this is no less true. Competition had always been fierce in this domain. But when Intel® introduced its own graphics program in the late 1990’s and took almost the entire low end of the graphics hardware market in the years that followed, it sharpened competition in that space even further. In fact, the competitive landscape in the graphics market promoted innovation. Some companies focused on a graphics program aimed at delivering 3D performance to differentiate themselves against standard on-board graphics. Other companies developed a graphics program aimed at delivering products that were optimized for more specific usage scenarios. Today, it is possible to find graphics hardware products that provide specialized capabilities for specific industries ranging from medical imaging, to air traffic control, and to areas where collaborative display walls are deployed. Almost all domains that benefit from advanced graphics capabilities have a corresponding product that best meets their requirements. Additionally, the range of graphics card options for workstations has grown considerably. This affords IT managers a selection of products that can be chosen based on the required feature sets for a given project. Such feature sets include 3D performance, image quality, long product life cycles, low power consumption, form-factor, and so forth.

When considered along with the even more ferocious competition that has taken place in the computer monitor space, driving the prices of monitors down considerably, the increased number of graphics options for control rooms have brought many new possibilities to IT managers. It’s therefore no surprise that the trend in high-reliability environments, where monitoring of real-time information is conducted, has been to deploy graphics hardware that supports the use of multiple displays. In environments where two monitors were once used, many operator stations have grown to now employ the use of three or four. Where four monitors were used before, some operator consoles have grown to make use of eight or even more displays per operator. Lower costs coupled with substantially better tools and greater software compatibility has made the use of multi-monitor workstations practically ubiquitous in energy control rooms.

This trend can be mostly attributed to the fact that the comfort level of operators monitoring vital data in multi-display environments has increased. The additional benefits to operators when working with a multi-display setup have become more evident. In such configuration, the content they are monitoring can be displayed in the most conducive manner, enabling them to maintain the best sense of scope within their applications. The ability to more easily retrieve data and interact with it has resulted in improved response time, increased user comfort, and an overall reduction in errors.

Competition in the Industry, which spawned a series of technological advancements, created an environment in which advanced graphics technology was widely available. It then took some time for users of many software applications to fully benefit from the productivity-enhancing, multi-display features of the hardware. However, that is no longer the case. Even in steady, slower-moving industrial and government areas, long-standing application content has evolved to provide users with more efficient ways to execute tasks, such as overviews of large grids, zooming into a specific area on a map, alarm management and, most importantly, supporting scalable multi-monitor capabilities that make it much easier for users to find appropriate application windows or sub-windows in their control systems.

Control Dynamics International Case Study – Reliant Energy
http://www.controldynamics.com/pdf/cdiEntexCaseStudy.pdf

“Multi-monitor technology with operator selected window assignments enables each operator to utilize four monitors, each to display system information and control. Special scripting provides a wide variety of window assignment configurations without redundant sets of windows for each monitor.”

Splitting the system from the user interface
Another key objective in control room design is to separate the user-interface components of the computer – such as the monitors, a keyboard, mouse, other USB devices and audio capabilities – from the main host computer. The typical reasons for this include the desire to increase security, the aim to reduce noise in the control room – by eliminating loud fans in workstations and servers, the aim to provide better humidity and temperature control in both the operator control room and in the room in which all the systems are housed, as well as the desire to increase redundancy, support, and maintenance options.

There is no shortage of competing technologies that help meet these objectives. Since graphics is the single most demanding element to remote due to bandwidth considerations, the ability to remotely power the user-interface from the host computer system is a subject that is intimately tied to graphics hardware.

One way to remotely power a computer from the operator workstation involves the use of hosted computing sessions exposed on network appliances. These appliances can be classical PCs or more passive pieces of equipment commonly known as Thin Clients. These appliances don’t run much information locally. They mainly serve to expose the actual computing session that is hosted on a server elsewhere on the network. Two main examples of software transports that expose these sessions are Microsoft® RDP and Citrix®.

There are many advantages associated with thin computing. These include the ability for user credentials to be managed from the appliance, for users to be able to log into several different servers or host systems, and the fact that the distance between the appliance and the host system can be virtually limitless.

There are also a few areas where the deployment of such a solution is less optimal. Possible disadvantages of using network appliances include the fact that they rely on network bandwidth. The protocols used typically limit the resolutions supported and, therefore, the number of displays that can be used. It is possible to publish either applications or the entire desktop. However, because of the protocols used, network appliances offer fewer options for configuring the desktop than those available with standard workstations. Some advances in graphics hardware and software now take better advantage of thin computing environments by providing multi-display graphics cards for the user-side appliance that consume very little power, and optional server-side desktop management software that allows the user to have access to a multi-monitor desktop that more resembles one powered by a standard workstation.

Another way to remote the host computer from the user environment involves the use of software algorithms to compress graphics signals, as well as those for other peripherals such as a keyboard and mouse, and to use the CPU of a user-side appliance – like a Thin Client or desktop PC – to regenerate the content and input/output (I/O) signals for operator interaction. In general, these approaches more closely associate the appliance and the server with one vendor and, as with thin computing, the transport of data is finite in bandwidth. This means that control room IT managers must pay close attention to the number of displays used and the overall burden placed on graphics. However, this approach can support virtually infinite distances between the operator and the host system, just as is the case with thin computing. The best graphics options for these solutions usually consist of products that provide maximum bandwidth for data transfer on the user appliance. Thus, PCI-Express x16 slots with compliant graphics cards are typically optimal.

Keyboard-video-mouse (KVM) Extenders take a completely different approach to splitting the user I/O from the host system. These all-physical setups offer a really key advantage in that they do not have any dependencies on software. They are completely passive in their role. These products usually operate by sending graphics, keyboard, mouse – and even sometimes audio signals – into a transmitter appliance on the host system side. All the data is compressed and then decompressed once it is received on the user-side appliance, connected to monitors, a keyboard, and a mouse. Despite its passive role, the robustness of this solution overcomes the key hurdles of distance and performance. It is costly to enable a remote solution that has support for up to four monitors at a considerable distance of 400 ft or more from the host system. But, more importantly, it is difficult to optimize the finite bandwidth used as the transportation mechanism between the host system and the user-side receiver to provide a truly responsive operator experience with energy management applications. Many solutions exhibit mouse latencies and support a very limited number of graphics resolutions.

Finally, bus extension offers yet another approach to splitting the user interface from the host system. Bus extension implies that the PCI or PCI-Express slot itself is physically included in an appliance far away from the host system. The strategic advantage of this approach is that the data sent between the host system and the user appliance is only bus data which, bandwidth-wise, is an order of magnitude smaller than graphics data. The key advantages of solutions based on this approach relate to simplicity and performance. Just as with KVM Extenders, they are used independently of specialized software applications and do not require the use of a special protocol. The only requirement is that the operating system supports the devices that are powered remotely from the computer. When using a product based on bus extension, mouse cursor and application performance is fluid. The user experience with an application is smooth even when working with a quad-monitor setup at very high resolutions.

When using any split-computing approach, at least some compromises need to be made. The inconveniences associated with solutions designed around bus extension include factors such as the fact that distance is typically finite – measured in hundreds of feet; specific transmission support requirements, such as standard multi-mode, fiber-optic cabling, may be applicable; and, as the appliance is an external peripheral to the host system, the link between the host system and the appliance must be preserved. A search term for solutions based on this technology is “remote graphics unit.”

Specifically for use in control rooms, since the host systems can be housed in the same buildings as the operators in most cases, bus extension generally offers the best blend of compromises for mission-critical monitoring at high performance.

Another advantage of bus extension is that it is one of the easiest solutions to combine with a collaborative display wall. In the event that a control room IT manager desires to display a copy of some, or even all, of the data displayed on an individual operator screen to a large, shared display wall at the front of the room, it is possible to do so using a host-side solution that has been designed around bus extension technology. For example, a host system with two free slots, either PCI, PCI-Express, or even one of each, could contain a transmitter card that would correspond with the user appliance. An additional graphics card inserted in the same system could then produce a clone of the content displayed on the user appliance. Rather than feed that cloned content into monitors directly, the content could be fed into the DVI or RGB capture ports of a display wall server. From that point, the display wall server can capture, scale, and re-constitute the data to the shared display wall in real time. More importantly, the data can be resized, positioned anywhere on the display wall surface, and present the exact information that is on the user station without any delays.

A more concrete example is an energy management room in which eight operators are facing a collaborative display wall that is comprised of ten rear-projection cubes. Each operator’s workstation is powered remotely by a host computer system situated in a secure room, equipped with a transmitter card that is powering the keyboard, mouse, speakers, and multiple monitors at the operator end. A quad-display graphics card is installed in each workstation along with the transmitter card and is cloning the content of the user appliance. The fourth output of the graphics card, which is producing a clone, is then fed into the capture port of the display wall server situated in the same room as the eight operator workstations. The display wall of ten cubes is displaying the content of the fourth monitor from each operator’s workstation. In this example, each cube can correspond to one full-screen of operator content. The two additional cubes can show content independently of what is shown on the user stations, including: a grid overview, various camera feeds, alarms, weather warnings, national security warnings, etc.

There are obviously countless configurations available but what’s most important is that it is now possible to implement control room collaboration without the extremely costly hardware and specialized software that had been required to do so in the past. Graphics support is scalable to individual control room requirements and, even more importantly, compatibility can be ensured across platforms developed by various software providers. There is no need for customization and heavy support costs. Rather, in cooperation with software vendors and/or project management firms, control room IT managers can enhance the basic infrastructure of their control rooms by further customizing their environments to best suite their needs.

The most important thing to remember when contemplating control room upgrades or new installations is that there are many ways to add productivity and value and interesting options abound. From basic multi-display graphics alternatives, to split-PC computing and control room collaboration, there are options. And options have value.

About the Authors
Samuel Recine is a Business Development Manager for critical decision systems based out of Matrox headquarters in Montreal, Quebec, Canada. He has been with Matrox for over 10 years and has extensive experience working with end users, integrators, VARs and other customers in areas including power generation, transmission and distribution, process control, mining, oil & gas, transportation systems, and pharmaceuticals. He can be reached at: srecine@matrox.com.

Liv Stewart has worked at Matrox Graphics Inc., in Montreal, Quebec, Canada for seven years. She is a representative of Matrox graphics hardware for use in energy management, distributed management, outage management and other mission-critical systems. She can be reached at: lstewart@matrox.com