AGI and GeoEye Gain Agility and Efficiency
by Bob Lozano, Appistry
Isn't it interesting that the data applications that today might rely most heavily on cloud computing are distinctly focused on Earth, but from the bird's eye view? That's the idea with clouds—you need to back up, sometimes miles, to get perspective.
Geospastial leaders all over the industry are doing just that, taking a few giant steps back and really looking at how they're doing business. They have seen that they've been asking amazingly talented algorithm engineers and developers to also take on parallel and distributed computing—and they're saying, "Hold it. This is inefficient. Let's figure out something else."
When these decision-makers stick to their core competencies, they accomplish two purposes:
- They let their developers focus on what they're really good at, creating a less complex, more efficient, and more agile working environment.
- They begin to use all the little pieces at their disposal, which means they begin to develop applications in a much more elegant, correct way—instead of the BIG, data-hungry way. They become mean and lean—specialized. Frugality often drives innovation. And the fabric-based cloud technology many geo companies are using has allowed for all of this to be easy—the drudgery computing that they could do, but that isn't a core competency, can be handled by the cloud (the Internet).
These are times for backing up to gain perspective, no doubt. Organizations are taking many steps in reverse to get the bird's eye view, the big picture. It is believed to be Socrates who said, "Man must rise above the clouds and look back upon the Earth. Only then can he truly understand the nature of things." That statement still rings true today.
Figure 2 Seoul-Inchon Airport, Seoul, South Korea, taken by GeoEye-1 on Oct. 29, 2008. Image courtesy of GeoEye.
GeoEye Uses the Cloud to Stick to Its Core
GeoEye (Dulles, Va.) is one company with this changing world view. They are a premier provider of commercial satellite imagery to the Department of Defense and intelligence communities. The company also provides satellite and aerial geospatial information to government intelligence, national security, and military organizations, as well as to commercial customers, including Google Earth.
Figure 3 This image of ocean ice flow around Vladavostok, Russia was taken on Jan. 5, 2009. This is Russia’s largest port city on the Pacific Ocean. It is situated at the head of the Golden Horn Bay, not far from Russia’s border with China and North Korea. Vladavostok is the home port of the Russian Pacific Fleet. Image is courtesy of GeoEye.
As part of the $500-million NextView contract from NGA awarded to GeoEye in 2004, GeoEye launched its GeoEye-1 satellite from Vandenberg Air Force Base in California in September 2008. From an orbit of 423 miles (681 kilometers) above Earth, GeoEye-1 has a ground resolution of 0.41 meters, or about 16 inches. The satellite also provides 1.65-meter resolution multispectral imagery. Due to current U.S. government licensing restrictions, non-U.S. government customers will have access to GeoEye-1 imagery that has been re-sampled to .5-meter ground resolution. GeoEye-1 is able to collect some 700,000 sq. kilometers of panchromatic imagery each day—an area equivalent to the size of Texas. It collects about half that in the multispectral mode. See Figures 1-3.
Talk about volumes of data and massive computational demands! From sharpening and compression to geo-correction, GeoEye's ingest applications perform a variety of processing steps using that massive amount of raw image data.
A launch like GeoEye-1 can change everything—and without the cloud, there would have been the daunting challenge of generating more capital fast for the multi-processor supercomputers necessary to muscle through this next-gen milestone.
Before GeoEye-1 was launched, GeoEye's leaders had taken those steps back and considered the cost and complexity issues associated with doing business as usual:
- Infrastructure costs - Between the initial price tag in the millions and recurring maintenance fees in the hundreds of thousands, purchasing high-end multi-processor servers just would not work.
- Hardware obsolescence - GeoEye's applications are typically deployed longer than the hardware on which they are originally designed to run, so system obsolescence—and the potential need to re-design a deployed application when vendor support for the hardware platform ends—was a real concern.
- Application development complexity and risks for multi-processor environments - Programming for symmetric multiprocessing (SMP) servers requires specialized development skills and prolonged development efforts. GeoEye could put its best mind to this task, but realized its skills were better spent focusing on the company's core value.
- Platform rigidity - Expensive, high-end servers require accurate, upfront forecasting of capability and scalability needs, a requirement that is neither flexible nor realistic.
- Innovation limitation - Once GeoEye's images are ingested, they are exploited by downstream applications that turn the images into actionable information. As with improving ingest applications, the high-end server approach creates barriers to GeoEye's ability to innovate new exploitation applications. Developers were forced to conceive of and design applications within the constraints of the infrastructure's limitations, stifling the company's agility.
GeoEye is using the cloud computing Appistry Enterprise Application Fabric (EAF) solution to build its next-generation platform for geospatial intelligence applications. EAF provides the following to GeoEye applications:
- Application-level fault tolerance - Fabric-based applications derive their dependability from the fabric itself, rather than from the hardware on which they run;b.
- Automated management - Application fabrics dynamically discover and assimilate new hardware and software, minimizing administrative and operational overhead; andc.
- Scale-out virtualization - Application fabrics can easily scale out across tens, hundreds, or even thousands of commodity computers, yet are viewed and managed as a single system by developers and administrators.
AGI Uses Cloud to Analyze the Situation in Space
Analytical Graphics, Inc. (AGI) also recently reaped the benefits of the cloud together with Appistry when it took its data-intensive libraries out of a singular group on the desktop and split them into multiple libraries. AGI develops commercial off-the-shelf (COTS) analysis software for land, sea, air, and space that is relied upon by the national security and space communities in more than 35,000 worldwide installations; it is a premier provider of space situational analysis.
For a vivid example: satellites can collide in space, creating debris of all sizes—and AGI handles the precise calculations to determine when that can happen. See Figures 4 and 5. Some of the calculations that are involved include satellite orbit determination, navigational accuracy calculations, and dilution of precision (calculating where the navigational data may have lost its accuracy in transfer). AGI and Appistry partnered to create custom and COTS applications based on AGI's Dynamic Geometry Library; these are easily scaled to meet the real-time needs of the applications.
Figure 4 AGI software assesses potential collisions of all space objects simultaneously. Image courtesy of Analytical Graphics, Inc.
Figure 5 This screen shot from an AGI viewer file shows a view of low Earth orbit satellites (green) and the space debris ring (red) from the Chinese satellite that they destroyed on January 11, 2007. The Xichang Space Center is also shown. This STK-generated image is courtesy of CSSI (www.centerforspace.com), and also appeared in our Summer 2007 issue in the article, Code Red: China’s Threat to EO Sats from Space Debris.
In lifting their application out of the desktop and separating it into more easily usable, more efficient pieces, AGI looked for ways to distribute this code among a range of machines to support it. The pieces are distributed across the fabric, which gives AGI's customers the ability to provision their systems only as needed, and also opens up an opportunity to reduce their reliance on those high-end multi-processor servers that they had previously needed.
Appistry was asked to provide a proof-of-concept, working with AGI tools to create the user interface, back-end processing, and necessary algorithms to be cloud-supported. AGI has found that their calculations now get back to the user more quickly, and their system boasts better fault-tolerance. If one machine fails, there's always a back-up.
AGI's customers are in the environment today where occasionally they may need the massive processor, but in the next five days they may need only a little of it. The fabric helps AGI's cust-omers conserve resources when it's not necessary to have them; it also knows when something needs more resources and it provides those right away. Using cloud computing allows delivery of the technology on any platform and in any form factor.
Next stop: space debris—a bigger universe for sure. Will more machines be needed for processing? Yes—and AGI can just plug them into the cloud.
The Bird's-Eye View Reveals
a Future of Opportunities
Key benefits of cloud computing include effortless scaling of the environment with no changes to the application's code base, enhanced reliability because hardware failures never impact completion of in-process application tasks, and reduced hardware and maintenance costs, with estimates showing that a single high-resolution image processing application running on cloud computing will save 88% in hardware and software acquisition costs and 58% in three-year recurring costs.
In addition, greater simplicity of development allows developers to focus on honing algorithms rather than on programming structure, and hardware obsolescence is eliminated with the machine-independent nature of cloud computing.
Using cloud computing, AGI and GeoEye may have created a "future-proof" environment. GeoEye and AGI not only can serve current customers better and more profitably, but also can grow their business in new directions with the only limitation being the companies' imaginations about how they can exploit their assets. With cloud computing, the sky's not even the limit. The limit hasn't been located yet, and we don't recommend breath-holding.
NOTE See also Next-Gen Mapping (Fall 2008), Nuages: Innovations in the Clouds: Innovations in the Clouds: A New Geo Marketplace.www.imagingnotes.com/archive.