Is cloud computing just another fad? Another technology solution looking for a problem? Brian Timoney thinks this fad is different because of “the variety of cloud-based services coming online in 2010 that directly address the problems that resource-constrained GIS shops face on a daily basis.”
IT Is Suffocating GIS
Ask average, ordinary stressed-out GIS managers about their day and the likely response will be a litany of grievances about servers, networks, licenses and unhappy users of Web applications. Note that most of the above have little to do with the traditional responsibilities of GIS departments such as maintaining positional accuracy of spatial data, spatial analysis and cartographic production. Instead, much energy is devoted to maintaining the complex choreography of connectivity between desktop applications, databases, servers and Web portals. With budgetary pressures showing little sign of abating, shops are forced to make difficult trade offs in managing their mix of IT and personnel expenses. The costs of this complexity are significant in ways that are both explicitly financial but also somewhat psychological, as industry veterans lose their enthusiasm under the burden of IT management responsibilities that hold little intrinsic interest.
Meeting Users Where They Live
Cloud services, whether they be disk storage, rentable servers or more recently, non-relational databases (e.g. Amazons SimpleDB and Googles App Engine), have been available for some time, yet a significant amount of know-how is necessary to get going. From comfort with the command-line to programming proficiency, you need chops to get the pieces and parts working together to create an environment productive enough to justify the effort. In short, a number of the headaches of in-house IT seem to have their direct corollaries in the cloud. To be truly effective, cloud solutions need to meet GIS users where they spend most of their time: their desktop.
Rent What You Need; Pay For Only What You Use
In the recent launch of its flagship FME 2010 product, Safe Software stated that the company was aiming to offer the software as a service using Amazons cloud infrastructure, which can be purchased in per-hour increments. The software-as-a-service (SaaS) model in the cloud counts among its success stories such familiar Internet offerings as Salesforce.com and Gmail. Whats compelling about such an SaaS offering in geospatial is that there are a number of "every once in a while" tasks for which the usual approach was to buy pricey extensions and/or additional licenses. Now, with a metered model, you can spin up FME instances on Amazon as your tasking warrants and pay for the additional functionality only when you need it.
Databases Without DBAs
While the shapefile chugs along well into its second decade, relational databases (RDBMS) are the preferred repository of spatial data in the enterprise. But along with all the performance and reliability gains of an RDBMS come management overhead. Managing tables and queries, slinging ad hoc SQL, and keeping the database "talking" to both desktop and server all require extra know-how. As mentioned above, both Amazon and Google offer non-relational databases that give up explicit relational structures in exchange for scalability and performance gains. As a plugin to the ArcMap desktop product, the most recent version of Arc2Earth (in beta) transforms Google App Engine into an "always-on" datastore with features directly editable in a familiar desktop environment. Further, Arc2Earth enables the direct publishing of your spatial data out of AppEngine in common formats such as KML, JSON, CSV, etc. Again, you pay for what you use - for both storage and CPU time. For many small- and medium-sized shops this combination of an online spatial database and publishing platform will yield monthly bills that will strain to exceed a few dollars.
Taming the Chaos of the Shared Drive
Once, while I was pitching a Web mapping solution, an enthusiastic listener inquired whether my proposed interface would work with the companys image server. "Why sure, if you can already publish your imagery via a standard such as WMS, we can stream that to the interface, no problem," I confidently stated. Later, when chatting with the department head, I mentioned the integration with the image server. "Image server? We dont have an image server. We have a server with a bunch of MrSid images on it." Indeed, the "tragedy of the commons" plays itself out on a daily basis on the shared drive that devolves into a dumping ground of data and folder names with nary a hint of metadata. WeoGeo, the "veteran" of cloud-based geospatial companies, has built a spatial data content management system on top of Amazons cloud. Offering more than simply a "not in-house" shared drive, content is cataloged so as to be both spatially searchable and available in multiple formats and projections. Deeper desktop integration via ArcMap tools is expected to be announced later this month.
New Business Models for Geospatial Functionality
Stepping back from specific features and functionality, its clear that a new model for delivering geospatial functionality is emerging. Vendors and users are no longer bound by one- or two-year release cycles, with disks arriving by mail plus periodic patches and bug fixes. Instead, the cloud enables vendors to roll out improvements on an ongoing basis without worrying about which users are on what version, since the software is controlled from a single location. Lower production costs combined with the already attractive infrastructure fees that Amazon, Google and Microsoft offer equal a potential total cost of ownership which would be an order of magnitude cheaper than the IT configurations found at many medium-sized GIS shops. Moving forward, it will be fascinating to see what tack entrenched vendors such as ESRI and Autodesk will take in offering users the significant benefits of the cloud while not unduly cannibalizing their existing revenue streams.
Long Live the GIS Analyst
Infrastructure-on-demand and pay-as-you go software that are tightly integrated with the desktop will positively reconfigure the default career paths in the geospatial industry. For many, higher salaries can only be found in the realms of project management, pure programmer/developers or systems administration - positions that take them far from the geographic interests that originally attracted them to the field. With the vast resources of the cloud readily integrated into everyday workflows, the mid-career GIS analyst can multiply her productivity (and, presumably, get a nice bump in salary) and deepen expertise in a world where models run in minutes instead of overnight, multi-scale cartographic representations are rendered as Web-ready tiles at the push of a button, and the huge volumes of data generated by LIDAR and the sensor web are stored and managed elegantly. Heady stuff to be sure, but the paradox of the cloud is about technology advancing so as not to be an end in itself, but rather taking its rightful place being simply in the service of creative spatial problem solving.