Sergej Khackimullin - Fotolia
Worker productivity is the mantra of enterprise IT, but it's often difficult to determine whether applications designed to boost efficiency are a help or a hindrance. Network data analysis, in the form of how network performance either supports or blocks business processes, can help.
Businesses have a number of tools to gather information. They can conduct time studies at the worker level, they can measure response time, and they can gather statistics that illuminate queuing and application response.
With this data and standard analytic software, it's possible to apply business intelligence principles to improve operations and productivity.
Applications, systems software, hardware platforms, network equipment and network services all generate performance statistics. We actually know more about how our IT/network infrastructure is working than we do about most business operations. That's because IT elements tend to keep their own statistics. The key to creating business intelligence out of performance data is to exploit the concept of workflow.
Gaining insight from network data analysis of workflow processes
A business activity can be characterized as a series of workflows that originate with a trigger event (a retail transaction, a request for analysis) and terminate with the return of the desired output. Businesses that have adopted one of the common enterprise architecture (EA) process models will already have a map of business processes that link with their IT applications for support. By combining information about the EA processes and their IT counterparts, it's possible to gain insight into how the business is running and how it might run better and more productively. Even where EA data isn't available, the portion of a business process that forms an IT workflow can yield valuable insights.
Two key metrics must be developed to drive business decisions from IT and network data. The first is application load. This is the rate at which business triggers request some type of processing. The second is application quality of experience (AQoE). This is the time required to return a response to that business request. The load data, then, is the sum of the transaction or request counts for a business' key core applications. AQoE is the response time measured at the point of worker interaction, be it a mobile device, terminal or computer.
App is slowing productivity if response times are too long
Where application load response times consume a large portion of worker hours, the application is likely constraining productivity by delaying needed information. This indicates a need to investigate the most cost-effective response; more on this below. In addition, look for indications that a worker is using the same application, or related applications, several times to obtain the information he or she needs to fulfill a request. This would suggest a new application user interface -- perhaps one that presents more information at once or one that displays different information -- would improve worker productivity.
To identify opportunities for new services, examine the employee/process side of the EA/IT boundary. In particular, look at business processes that occur offline before they move online and, thus, require IT support. If workers are accessing information from files, reports or other offline options, moving this information online could shorten the response time needed to provide business results and improve worker efficiency.
Where EA data isn't available to assess how offline information resources are used, try tracking the reports created by the core applications to see how they're used. Many companies have perpetuated the use of reports and files simply out of habit, even when better online processes are available. Track report distributions and analyze usage to determine whether better online services would supplant offline resources and bolster worker productivity.
Network data analysis of IT and resources
Taking the network data analysis in the other direction -- toward IT and network resources rather than from the worker/processor side -- can also pay dividends. There are two ways to approach this: one from the worker/AQoE side and the other from the transaction response perspective. Where business process data is available, AQoE analysis produces better results faster. If no such data is available, then examine transaction response.
The AQoE model takes business processes as the starting point and arranges IT transactions in order based on how they're used. By using application response time measurements, it's possible to see how much time is devoted to IT activity. The applications should then be ranked by their contribution to total process delay and examined to identify bottlenecks.
It's important to look at network delay, queuing delay at the application level, and application processing time as three independent processes. Network monitoring probes -- placed at select points in the traffic flow and designed to use deep inspection to identify specific application behavior -- will produce the best results in analyzing network performance. These probes can produce data graphs over time, which can then be correlated with similar time-based information from applications and also with patterns of work through the day. This approach will generate a more accurate picture of how response time issues are divided between IT and the network.
Remember that understanding workflow is crucial if you are going to be successful in using performance data to reap business benefits. If you don't have organized information on how IT relates to business processes and how information moves among applications and to/from workers, it will be difficult to establish the context that's essential for any analysis of performance data. Take some time up front to get this right. Allow workflows to govern how performance statistics are organized and it will pay large dividends.
What's next for business intelligence?
Designing a business intelligence system for business use