Defining Real Time

You have had the buildup, now here is the pay off. This is my formal definition of real time.

If the time between when the information is created and when it is available for analysis is imperceptible, then it can be considered real time.

Cool huh? The fun part comes when you tackle the word imperceptible. What does that even mean? There is a lot of latitude in the word “imperceptible”. The reason the word imperceptible is in the definition is because what is considered real time will vary by organization. The question that you have to ask is, what is the interval between when the data is generated, and when someone will notice that it is not there.

Making the definition this flexible rises out of the reality that different organizations analyze their data at different rates. I have worked in organizations where people are pulling data all day and executives really do need the most up to date information. Other organizations have a slower decision cycle where they have fewer analysts and it takes longer between request. It might be so long that you could get away with a micro batch process where you have a batch process that executes more frequently than overnight and with less data, although I do not recommend that. I always tell clients that real time can be five minutes or less.

As a practical matter, the technologies required to implement real time analytics usually only work at one speed. As fast as possible. However, when you are talking to executives about implementing real time systems, you have to ease them into it. The concept of getting data “right now” will blow their minds and they will resist. They will resist because they have absolutely no frame of reference for shooting data around that fast in the organization. If you can just talk in terms of “faster”, it is a much easier conversation to have, and then you can wow them later with data that does the Kessel run in less than 12 parsecs.