Next Generation Storage Architectures for Imaging

By Paul Rutherford

Increasing customer demands for better, faster and higher-resolution geospatial solutions - from governmental bodies to online earth-mapping - are imposing seemingly insurmountable demands on many GIS systems, which have, until recently, relied almost exclusively on conventional storage area network (SAN) and network attached storage (NAS) systems. Those systems are gradually hitting performance limitations, which are paralyzing GIS companies by limiting productivity, image resolution and the ability to keep up with client demand.

The answer, as two large international GIS companies have discovered, may well be "scale-out NAS," which has not only helped them conquer data processing performance, but has enabled them to increase output and decrease project turnaround times, as well as significantly expand their business opportunities.

So what are NAS and "scale-out NAS"? Traditional NAS consists of an intelligent compute server or "head" in front of any number of disks. That means at scale you either a) reach the limitations of the head to serve read/write requests for data on the disks, or b) must purchase an entirely new - and very costly - system to keep pace with your data growth. When you consider that traditional NAS volumes are limited to 16 TB or less, an NAS system may have tens or even hundreds of volumes within a single appliance, fragmenting critical data and slowing access. Add to this challenge the process of bringing additional, separate systems online to manage data growth and you're left with fragmented data that cannot be effectively accessed or leveraged to drive business operations. In short, traditional NAS creates complexity as data grow and the organizational realities of managing this complexity are simply not sustainable.
Isilon scale-out storage creates a single, easily expandable file system that accelerates application performance and reduces costs for geospatical solutions companies. This image shows a rack of Isilon's S-Series storage devices.

On the other hand, scale-out NAS is designed to keep pace with both today's data growth, as well as tomorrow's unforeseen challenges. Scale-out NAS eliminates the key challenges of the traditional NAS model by creating a highly flexible, on-demand storage architecture in which each hardware node is an intelligent peer, meaning there is no head limiting access to disks. Each node houses CPU, disk, networking and software intelligence to form a single, easily expandable file system in which all data are unified, immediately accessible and available. By combining the volume manager, file system and RAID (redundant array of independent disks) into a single operating system and single point of management, scale-out NAS departs dramatically from traditional storage, delivering a central storage resource which can be scaled on-demand by simply adding nodes to the cluster to meet data growth and performance demands, while reducing costs.

North West Group
Alberta, Canada-based North West Group is a leading aerial data acquisition company that provides high quality aerial mapping and related spatial data services, including digital remote sensing, digital aerial photography, LiDAR (Light Detection and Ranging) and related services in North America and international markets. A significant portion of the data it generates comes from the widespread deployment of its six aircraft that are outfitted with third-generation, large-format ADS-80 airborne digital sensors, which capture extremely high-resolution digital imagery at a rate of roughly one Terabyte (TB) an hour. The data, once returned to North West Group's data center, are stored and processed in a variety of ways depending on each client's formatting needs.

"A lot of the data we create are for government agencies," said North West Group's vice president of technology, John Welter. "So our images go into the public domain and end up in big maps, virtual earths and a variety of other portals."

In business since 1968, the company hit data-capture performance limitations with its clustered NAS system in October 2008. As its client projects became increasingly more sophisticated and the demand for higher resolution imagery grew, North West Group needed a high-performance, highly scalable NAS solution that could process higher-resolution data more efficiently and turn around projects more quickly.

"Our previous storage system couldn't handle our multi-tiered workflow, creating performance bottlenecks that impacted our entire operation," said Welter. "It placed a huge burden on our IT infrastructure. The problem with that system was that we reached its performance limit. Once we hit 350-400 megabytes (MB) per second, it wouldn't go any faster. It didn't matter if we added more disk space or additional clusters; it had simply reached its limit. We needed something that would scale out and keep pace with our work load."

Typically, once data acquisition from the aircraft is complete, the raw imagery is delivered to a company data center. From raw data, the images move through various photogrammetric processing steps, including pixel manipulation and stitching/seaming, all requiring the accurate, efficient and reliable storage, management and manipulation of large amounts of data.

For North West Group, those data move from level zero to level two, where the captured images, produced in long strips of imagery in relation to how the aircraft sweep the landscape, are sewn together to produce an image mosaic. Once the image has been sewn together, it reaches level three, where it is prepared for final delivery - edited to size and image type - per each client's specific needs.

With its previous clustered NAS system, North West Group was forced to manage recurring data bottlenecks, requiring constant, round-the-clock manual vigilance and management to keep data moving through the workflow.

"We could never break that 350-400 MB per second throughput ceiling," said Welter. "If the cluster was busy doing the raw-to-level-three data processing, then the human interactive steps would be slow as well. We'd have to have some poor soul scurrying back and forth 24 hours a day moving data."

The same problems were hitting Sanborn, a Colorado Springs, CO-based comprehensive, end-to-end geospatial solutions company that provides geographic and location-based information through custom applications including systems integration, spatial analysis and modeling.

With its precise aerial imagery and data-gathering abilities, the performance and scalability demands of Sanborn's image processing workflow overwhelmed its previous NAS system - limiting processing performance and workflow productivity.

"We were witnessing government and private sector demand for our geospatial solutions rapidly increase, which was great for business but a big headache for our previous NAS system. We had to make a change to take advantage of market demand," said Doyle Heisler, director of IT for Sanborn.

With multiple planes, each equipped with an extremely high-resolution DMC camera system producing hundreds of thousands of images per year, Sanborn needed a high performance storage system that could accommodate increasing customer demand, while simultaneously reducing the unnecessary storage management and escalating costs of maintaining its previous NAS system.

Both Sanborn and North West Group found relief and a significant improvement on ROI, with the deployment of scale-out NAS systems provided by Isilon. The use of scale-out NAS unified each company's image processing workflow onto a single storage resource capable of scaling performance and capacity on-demand to meet the unpredictable needs of their data-intensive operations. Sanborn and North West Group were both then able to complete more projects in less time, while reducing storage management to less than one full-time employee, driving increased operating leverage across their organization, and enabling each company to take on previously unattainable levels of new business.

Specifically, after installing a scale-out NAS system, North West Group achieved nearly double its prior throughput for both raw data and image processing. The single storage pool, available to all workstations, increased efficiency, reduced management and enabled the company to easily stay ahead of customer demand and pursue new business opportunities.

North West Group runs 160 HPC nodes in its data center and 35 production machines. When data arrive, they are copied directly onto the storage cluster, where Ethernet-attached, production floor workstations can access them for interactive processing. "Immediately, we almost doubled our throughput and got to 800 MB per second," said Welter.

System management was also dramatically reduced. "One of the nice benefits of the scale-out NAS solution," said Welter, "is that we probably average one to two hours a week of storage system management, total."

For many GIS businesses, customers have been on an upward climb for higher resolution data and faster turnaround times. "This system," offered Welter, "enabled us to accelerate our process, ultimately meeting or exceeding customer demand."

Welter cited a recent example. In 2009, North West Group aerially imaged the entire state of California and had 30 days to process and deliver product. "We got it done in 21 days," he said. "We could never have gotten that level of performance without this system. With our previous system, it would have been nearly impossible to hit that 30-day turnaround time and the data management would have been a big issue."

With the installation of a scale-out NAS system, Sanborn had similar results. "It was clear we needed a truly scalable solution for initial data storage," said Heisler. "With our old system, we had created a number of different file volumes. By deploying Isilon IQ, we not only created a single shared pool of storage for all image processing, but we also dramatically increased image processing speeds by 533 percent, enabling us to complete projects in much less time and open our doors to more new business."

With scale-out NAS, Sanborn now processes 8,000 to 10,000 images per day, compared with less than 2,000 a day prior to implementing the new system.

Both companies now rely heavily on the capabilities and reliability of scale-out NAS. They are also able to deploy multiple aircraft for increased data-gathering per project, enabling them to efficiently hit the weather windows critical to completing aerial imaging missions.

As more GIS companies hit performance ceilings with conventional NAS and SAN systems, scale-out NAS systems will become increasingly attractive as they provide the ability to seamlessly boost data-processing performance, decrease manual maintenance and project turnaround times, and ultimately grow revenue.

Published Friday, May 28th, 2010

Written by Paul Rutherford

If you liked this article subscribe to our bimonthly newsletter...stay informed on the latest geospatial technology

Sign up

© 2017 Directions Media. All Rights Reserved.