24 March 2009

Alerts & Exceptions using BI as part of Business Performance Monitoring

Today we take a look at the promise of BI: Actionable information.

Usually, we set up a complex set of metrics to measure the things that business would have us measure. These could be anything from OH&S incident issues, on-time delivery, average collection period, financial leading and lagging indicators and the performance indicators of where the business is. These are tactical and strategic initiatives.
However, Business Intelligence can also be used to take action immediately when something that crosses preset or dynamic thresholds, happens. A simple example might be for a bank, where each branch for each region has a simple threshold for deposits taken at the branch. Let us say this figure is set to $50,000 for Branch 00-11-22. If this threshold is exceeded, a text message will be sent to the financial adviser / risk & compliance officer of the particular branch, notifying them within 60 seconds of the transaction hitting the system.

The BI Tool can monitor the message queues from the bank transaction system (like MQ Series, JMS, etc.) or it can directly interrogate the OLTP datasource, looking for any transactions since the last time the check was done, for any records that meet the selection criteria. For this to work against real-time mission-critical data sources, it is imperative to have fast query performance that will have minuscule effect on the source systems, or if there is a near-real-time data warehouse that can serve as a datasource, the effect is minimized. This is achieved with a pre-written, tuned query, which is predictable.

This simple diagram shows the processes that should be involved for monitoring exceptions, and KPI's for that matter too.

Once one or more records have been identified, a lookup process can occur to see where these records' information needs to be sent (destination).

Once we have the composition of the information to be sent and the destination (an email SMTP server, a SMS Gateway, a SOA enabled application server like SAP , a web destination, etc.), the action designed can be taken.

So from this we can see that BI can be used as a business process monitoring tool as well as a business performance management tool too. The example shown here is simple, but with a near-decent rules engine (or make a call to a SOA enabled service), the net result is that the business is aware of a thing / event "happening" in their business in almost real-time, which is a whole lot better than being told via a report the following week that there was a transaction that could possibly have been a fraud or part of a money laundering operation.

The latency of information in mission-critical environments is inversely proportional to its value, thus for maximum value in these types of scenarios, real-time information (or as close to it as possible) yields maximum value to the business.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.