Embedded-device makers have a unique opportunity to leverage revolutionary new approaches to a longstanding problem: providing effective technical support for devices installed in remote locations. For makers of Internet infrastructure equipment, the news is particularly promising. Traditionally makers of routers, switches, cellular base-stations and the like have relied on network-management tools to identify problem equipment, then on traditional call center processes or a field dispatch service to fix it fast. The majority of problems today are still resolved through these laborious, surprisingly unautomated processes.
The service-quality management model divides the life cycle of a typical technical-support problem into five phases. Well-established management tools have optimized the first three of them: early detection, isolation and queuing. Yet the final two stages-diagnosis and resolution-can account for up to 70 percent of the problem's life cycle and are still handled largely by traditional processes.
Internet infrastructure vendors feel the limitations of present-day tech-support approaches perhaps more than any other market sector. Their equipment must perform at the highest levels of availability, and yet problem resolution involves analysis of tremendous volumes of data, usually locked beyond reach at the customer site or not captured because it wasn't thought to be of consequence.
Take the example of a wireless equipment vendor. Its support problems will tend to fall into one of three categories: configuration issues (the customer changed the configuration yesterday without alerting the vendor); documentation problems (a previous change was not recorded, and thus is affecting present function); and technical matters. Technical problems that recur frequently are easily categorized by their unique data parameters and readily resolved. But it's typically the intermittent, sporadic malfunctions that are the hardest to track.
For example, in addition to the more common configuration issues, vendors of wireless equipment also encounter occasional, difficult-to-diagnose technical problems such as weather-related failures, temporal obstructions-for example, when a crane for a time obscures line of sight between a basestation and a transmission tower-and gradual obstructions, for example a tree growing in the line of sight and over time diminishing signal quality or strength. These vendors need a way to quickly pinpoint more common problems, and analyze deeper data automatically to narrow down the possibilities-without having to send a support rep to the site, only to discover the source of the issue could have been easily deduced.
Exploring the impact
However, the vendor may only store data for a short period, not for the year or more that would be necessary for review. Vendors aren't able to explore the impact of new variables not formerly thought to have contributed to the failure.
In all cases, using historical data to make comparisons is key to solving problems, while automating the collection and analysis processes provides the speed and efficiency needed.
The sensitive matter of customers' firewalls is a major roadblock to the vendor capturing all the data it needs on an ongoing basis. Access to the data often meansthe customer must approve a direct connection-that is, open the firewall.
New models are emerging to provide remote technical-support solutions for embedded devices, including network infrastructure devices and other non-PC elements containing a computer chip. Recently introduced solutions are bringing the functionality of so-called "e-support" solutions for PCs-meaning full-remote technical support over a network connection-for specific embedded-device application categories, including Internet infrastructure equipment.
A new model of remote technical support for embedded devices, which targets network infrastructure equipment in its first implementation, addresses the three most significant barriers to effective remote support: collecting and analyzing the right data automatically, without creating system overhead or overwhelming staff with complicated displays; providing direct access without breaching network security; giving those who maintain network systems control over their data without the burden of collecting it.
This new approach uses three software components. First, site servers collect, store and transmit device data to the support organization and are deployed at customer sites, inside the private network. Second, vendor servers receive and manage device data and provide intelligent analysis and graphical visualization of the collected data, which can be viewed using a standard Web browser. They are deployed at the vendor location. Third, communication servers enable secure online communications between site and vendor servers and are deployed at the vendor location.
The new model works as follows: When customers discover device problems that they can't solve, they instruct the site server to automatically collect vendor-defined support data and deliver the packet to the vendor in the form of a problem report. This is a one-click process requiring no manual data collection. Customers control which data is sent, and all information is delivered via the infrastructure in a highly secure link. Optionally it can be e-mailed, packaged onto a disk or sent in whatever way the customer chooses.
When the vendor server receives the problem report, it generates a trouble ticket and e-mails the support representative. When reps get these e-mails, they simply click on the URL included in the message, which takes them to a Web page showing the analyzed data. A "summary" display highlights warnings, configuration changes, thresholds and other predefined parameters. Reps can then dive deeper into any element of the collected data by clicking on areas of the problem summary and viewing additional details.
To generate this analysis, the vendor server passes the raw data through a series of vendor-defined parameters containing mathematical functions, rules, statistical and trend analysis, and change detection, as well as any customer-defined analysis. The analyzed data then passes to a visualization module where results are presented in predefined, configurable textual and graphical displays.
If support reps require additional data, they can generate a new set of collection rules and send it back to the customer at any time. The new rules automatically install on the site server, if authorized by the customer, and can either send new data back immediately or collect new data over a period of time-replacing or augmenting existing collection rules.
The approach described above lets network infrastructure vendors overcome drawbacks of present approaches to remote technical support and makes the overall technical support function a time-savings for customers, a revenue advantage for their companies, and an availability boost for both.