Data Crunchers
Since it was founded in 1994, DMTI Spatial has facilitated innovative location intelligence services to Global 2000 companies and government agencies, including Canadian Tire, Rogers Communications, DHL, Union Gas Ltd. and Canada’s Department of National Defense.
By culling aerial photographs, satellite imagery and other locational data from thousands of government and industry sources, DMTI develops a comprehensive picture of a company’s customer locations. While those services are important, DMTI’s sweet spot is its ability to link locational data to other analytical systems to help clients such as DHL and ConocoPhillips Canada understand their own customers better and target products and services to them more effectively.
The locational generation process “comes to a head in an area we call master address management,” says John Fisher, chairman and chief technology officer for the privately held company in Markham, Ontario. External customer information maintained by DMTI’s clients is often contained in business “silos” and typically isn’t shared across a client’s enterprise, says Fisher. “The piece that unifies them is the address — the address of the customer and the address of the service location. We tease out that address location, we validate those are real addresses, and we link all of the information from that address and maintain it over time.”
The other aspect of DMTI Spatial’s business is housed in its reference databases, says Fisher. “We keep deep spatial data for all the streets, routing and logistics information.” The information is gathered via satellite and other imagery, and includes data on elevation, postal geography, census geography, lakes and railways.
For telecommunications industry customers such as Rogers Communications, which supplies multifaceted communications such as wire line, wireless, Internet and video, DMTI Spatial provides a more consolidated view of the services used by customers and improves customer service responsiveness by determining a customer’s location and the services offered in their region, says Fisher. For retail customers such as Canadian Tire, DMTI can provide store location analysis.
DMTI can also scrutinize locational data to help financial services customers analyze mortgage insurance requests, along with the information and risk pertaining to property values, all within 20 seconds, says Fisher.
All of this number crunching and imagery requires a massive amount of computing power and storage capacity. DMTI stores a whopping 25 to 30 terabytes of locational information from thousands of disparate sources, including federal, provincial and local governments, says Eugene Khaitov, the company’s IT manager. To handle such large volumes of data, the company installed an IBM TotalStorage DS4300 storage area network (SAN) system three years ago and complemented it with two IBM System Storage DS4700 SANs in recent months, says Khaitov.
Selecting a set of SANs over iSCSI and network-attached storage (NAS) environments was “an obvious choice” for DMTI, says Khaitov. For starters, the company’s primary Oracle 10g database is 10TB in size. “For huge input/output, we need a very sophisticated storage system,” he says. Plus, SAN environments are faster than iSCSI and NAS for database access and are best suited for DMTI’s dynamic growth. “We didn’t need to get huge storage up front,” Khaitov says. “We got small and upgraded on an as-needed basis.”
DMTI executives also found that IBM’s SAN systems were considerably more flexible in terms of meeting DMTI’s unique business processes. For instance, when DMTI downloads satellite imagery for locational use, the information is formatted in a given size, but, Fisher says, “it’s only a fraction of what it becomes during processing,” in which the file sizes typically swell by five to 10 times before they are condensed in the final production stage. “So we need that flexibility to handle those big bursts in storage requirements,” he says.
For the type of spatial data that DMTI utilizes, a SAN environment “probably makes the most sense” among data storage environments, says David Sonnen, lead geospatial industry analyst for IDC, a market research firm in Framingham, Mass. By contrast, a relational database or a data warehouse would probably make more sense for DMTI customers, who need to access geospatial data more frequently, Sonnen adds.
Lean and Green
For the past three years, DMTI has relied exclusively on its IBM SAN environment to store all of its spatial images and other data. The approach fits well with the company’s strategy of supporting its 80 employees and its customers with a sparse IT staff of just three people, including Khaitov. “We run a lean, mean IT organization, so we try to get maximum benefits out of our IT resources,” says Fisher.
DMTI’s tapered technical approach is also evident in its server environment. The company has been using IBM BladeCenter chassis for the past three years, which consume far less power than conventional rack servers, says Khaitov.
When Khaitov joined DMTI four years ago, he directed the company’s adoption of desktop virtualization, which yielded such positive results that the company decided to virtualize its Windows 2003 Server environment. Last year, using VMware, DMTI installed a single virtual server — a 16-core IBM System x3850 M2 machine — and plans to deploy another in the next few months, says Khaitov.
DMTI’s server strategy allows the company to operate lean and green. By moving to a single virtual IBM server, DMTI avoided an expansion of its 240-volt power plant and roughly $100,000 in capital equipment costs for additional cooling and server room capacity, says Fisher.
DMTI’s effective approach to leveraging IT has helped it become the leading geospatial solutions provider in Canada, says IDC’s Sonnen. The company’s ability to gather locational data from so many sources and then apply it for analytical purposes, says Sonnen, “has made them the primary source for mapping data in Canada.”
