A key catchphrase when defining computer system requirements is "timely."

Back in the day, "timely" often meant weekly, monthly or even longer cycles, such as quarterly.

Today, however, we are faced with much tighter time frames, including the holy grail of "real time." Most folks assume real time to be instantaneous, but in fact, there are many ways to interpret this concept.

The most common area where we see the phrase "real time" thrown around is in the area of reporting. Decision-makers want to know what is happening in their shops so they can adjust operating patterns, resource utilization and the like. The sooner they have accurate information, the theory goes, the sooner they can act.

The quandary many run into is that the systems that are used to support business operations are often not suited well to providing informational reports. Sometimes multiple operational systems don’t share the necessary data, but oftentimes a single system just isn’t structured properly to support both operational and reporting requirements. To a layman this might seem nonsensical, but it is a common problem.

As such, organizations develop secondary systems to support reporting requirements. The secondary system is fed by the operational system, and data is restructured to be easier from which to generate meaningful reports.

Herein is the source of the real-time problem. It takes time to extract data, transfer it to the reporting system and restructure. This includes processing time, which can be significant, but also network time, especially if the operational systems are in a different location from the reporting system. For example, operational systems could be out in the field, such as in the case of a retail store, while the reporting system is at company headquarters.

Network latency is roughly measured at 1 millisecond per hundred miles. This means that data going to the mainland takes at least 25 milliseconds. While this might not seem like much, network latency can adversely affect system perform severely.

Latency notwithstanding, advances in processor capabilities known as ETL (extract, transform, load) have greatly improved them. Instead of measuring the ETL process in weeks or even months, we now look at it in terms of minutes or hours (days at worst).

As such, many folks claim to have this process down to real time. At least, that’s what they tell their bosses. As described above, however, this is not an instantaneous process. No matter what, some degree of processing must take place, not to mention network transfer time. In contemporary systems, we’ve seen real time be applied to time frames of mere seconds up to 20 hours. So real time can be anything less than one full day.