For the past twenty years, application architects have debated whether to deploy software applications on clients or servers or some combination of the two. This debate has spanned a half-dozen different types of computing platforms, from mainframes and minicomputers in the 1970s and 1980s to desktop and client/server systems in the 1990s, and to the Web and Web services in the 2000s. Today, application architects now have a new platform upon which to wage their perennial debate: mobile devices.
Mobile devices resurrect the notion of "fat client" that had largely disappeared with the advent of Web-based computing. Fat clients generally offer fast performance and superior user graphical interfaces, while Web-based applications--which execute application code on a remote server--generally simplify development, deployment, and administration.
Over time, fat and thin client architectures generally blur into distributed architectures where application processing is spread across clients, application servers, and databases. In fact, a good portion of many Web-based applications today execute on the client via browser-based Java applets, Active X controls, DHTML and AJAX, or Flash plug-ins, making so-called Web "thin" clients a tad "heavier" than is generally known.
As a new application platform, mobile devices pose a stark architectural choice for organizations that want to deploy mobile BI applications. They can either build exquisite-looking, native BI applications for each mobile device, or they can build a single browser-based application that runs the same on all devices.
During the past year, every BI vendor has had an intense, internal debate about which mobile BI architecture to adopt. Some vendors, such as MicroStrategy and Mellmo (makers of Roambi), believe that users will demand feature-rich, device-specific mobile applications, while others, such as Transpara and QlikTech, are betting that browser-based mobile applications will eventually steal the day. Many others, including IBM Cognos, have decided to hedge their bets and produce offerings in both camps.
User organizations need to carefully evaluate the mobile BI architecture adopted by their mobile BI vendors. They need to understand the ramifications that their vendors' chosen architecture will have on ease of use, adoption, deployment times, maintenance, and total cost of ownership, among other things.
Let's take a look at the two mobile BI architectures and then examine their hybrid offshoots.
Native Mobile Applications
There are three main advantages to running native mobile applications: 1) performance 2) user experience and 3) offline access. Performance is exceptional with native mobile applications because the data that comprises a report or dashboard is downloaded and stored on the device. Fast performance makes applications very responsive and highly interactive.
The user experience of native applications is also superior. This is due in part to fast performance but also because native applications exploit unique features and functions inherent in the devices, such as touch screens that support hand gestures or an accelerometer that lets users rotate the device to switch between a portrait and landscape view. Anyone who has ever used an Apple iPhone or iPad understands the unique experiences these devices make possible. Having tasted paradise, most don't want to go back.
Finally, because native applications store data locally, they work even when the device is not connected to a network. Offline access is critical for executives who spend a lot of time in the air or salespeople whose districts don't offer reliable network connections, or plant managers who might lose connectivity in the bowels of a manufacturing plant.
Conversely, browser-based mobile BI applications have three unique advantages: 1) portability 2) data consistency and 3) security.
In a browser-based mobile BI applicaiton, the application and data reside on a remote server, which means that nothing is downloaded to the device. This prevents data from getting out of sync or the creation of analytic silos. Individuals all access the same server-side reports and data. Additionally, it is easier to enforce security since all data, code, and passwords are managed centrally. If a device is lost or stolen, hackers can't walk off with sensitive corporate data.
Downsides and Hybrid Solutions
Interestingly, the advantages of each architectural approach represent the downsides of the other approach. Not surprisingly, given the market potential for mobile BI, vendors are quickly plugging the gaps in their defenses and delivering hybrid solutions that maximize the benefits of their preferred architectural approach and minimize the downsides.
Hybrid Native Applications. BI vendors that have opted to ride the native mobile BI application wave have already made great strides in improving their applications' portability, data consistency, and security. For example, some now offer universal application templates that can be compiled into the native languages of all major devices without changes. This means developers only have to create an application once to run it on any device.
To improve data consistency, some vendors now configure native mobile BI applications to download fresh data upon startup or on demand. And they've given remote administrators the ability to selectively wipe device disks if a device is lost or stolen to prevent unauthorized access to sensitive data and require users to log in when the application is used offline.
Hybrid Web Applications. Browser-based mobile BI developers have created workarounds to improve the performance and usability of their applications. For instance, some now cache small amounts of data on the device (e.g., a report) to improve application responsiveness. Others are extending their browser-applications in the short-term by writing to a mobile device's application programming interface (API), which enable browser-based applications to exploit some unique mobile features, such as touch screens, accelerometers, and gyroscopes. Of course, this also creates a new version of the Web-based application that developers have to maintain. In the long-term, these vendors are betting that HTML5 will deliver these same features using industry standard HTML.
It's likely that the future of mobile BI is some form of distributed computing comprised of of the hybrid applications described above. But whatever the form, it behooves organizations to understand the architectural choice that they or their mobile BI vendor have made and its ramifications on application usability and internal resources.
Posted April 21, 2011 3:43 PM
Permalink | 2 Comments |