Technology Maturation Created the Cloud
When the computing age took off in the 1960s, powerful mainframe computers ran bulk data processes inside large organizations. At that same time, several companies realized they could take advantage of these large-scale systems by selling chunks of processing time to smaller companies that could not otherwise afford such a computer system on their own. Thus was born the concept of "time-sharing" services.
The idea was that access to one behemoth computer could be distributed and shared by multiple users simultaneously running different programs. Enabling many concurrent interactions allowed for maximum efficiency of the system, driving overall costs down. Time-sharing occurred in an off-site environment, complete with storage and various software programs. Users were simply charged for time used on the system, based on connection time, CPU time and disk storage. Sound familiar?
The emergence of the PC in the 1980s gave rise to the independent, fully-detached computer system. However, the explosion of the Internet in the 1990s was the catalyst for individuals and organizations becoming interconnected. This re-defeated the idea of system isolation, as it ushered back in the notion of shared computer resources; this time reincarnated as millions of users accessing solutions spread over vast server farms. All of this led to what we now know as "the cloud," and the reinvention of the "time-share" model.
The cloud as we know it today offers on-demand, network access to robust and rapidly scalable shared computer resources. Service is typically offered on a pay-as-you-go subscription fee. The cloud is based on the idea of self-service over the Internet, where subscribers with a remote login capability can rapidly provision system resources to stand up their solutions. The entire computing and storage solution is owned and managed by the cloud provider, relieving the burden of IT infrastructure ownership and operational costs. For many organizations, the cloud offers the simplicity of standing up solutions quickly without making huge investments.
Geospatial Industry Maturations and Growing Infrastructure Requirements
Computer systems, software applications and almost all supporting technology have matured at a rapid pace over recent decades. All aspects of computing, including processor, network, storage and memory capabilities have seen exponential increases in power and efficiency. Databases and application servers are extremely proficient at managing and delivering massive amounts of data and information to millions of users.
Geospatial data and associated systems for management, analysis, processing and delivery have also seen extraordinary advances over the past decades. Only 15 years ago, the best imagery collected was a SPOT or Landsat scene at 15 to 30 meter resolution or an aerial at 2 to 3 feet. A typical Landsat scene was only 2MB, but any processing on imagery of that size could take hours and max out a standard computer system. In general, data collection was infrequent, processing took long periods of time, and data were delivered via snail mail on CDs, if not tape.
Fast forward to today, where geospatial data are collected everywhere, constantly. GeoEye delivers .5 meter, high resolution satellite imagery in a matter of days. The dataset may be gigabytes in size but you can still download it over the Internet. Running geospatial processes on that scene can now take mere minutes or seconds.
But there are still issues that confront us today. Some are new, but some are the same old issues with new twists to them. Geospatial data are larger than ever, both in file size and quantity. And so are our rapidly expanding archives of data, as organizations are collecting more data, more often. We now use so many types of data in our processes, including high resolution satellite imagery, numerous feature datasets, laser scanned point clouds, sub-centimeter GPS points and more. Handling of geographic data has historically been computationally intensive, and algorithms associated with processing data have grown more complex.
In addition, organizations now expect to batch process multiple scenes at one time, or even parallel process them across a networked group of computers. This requires extensive processing power. The advent of location-based services and map services means we have to put more power behind the delivery of data to our end users.
Almost all organizations managing geospatial data are painfully schooled in the variability of demand for geospatial data, particularly on fluctuations in public demand related to isolated, disastrous events. Take, for example, a county government that is struck by a flood event. For a given period of time, the data management, delivery and storage requirements will soar. When the disaster is over, operations will return to business as usual. The county government must provide a means to accommodate the sudden strain on infrastructure resources, and also the means to return to normal daily operations.
From an IT infrastructure perspective, this becomes a management nightmare. The more data we get and the more users we serve, the more infrastructure is needed. Plus, we need the means to catalog and make sense of all of these data, and disseminate them efficiently to users across our organizations.
This all sounds quite daunting, but for some software vendors, these "new" requirements are converging to make things more exciting in the geospatial industry.
Ed. note: Part 2 of this article will appear next Wednesday, Aug. 18.