“This is the 19th day of my participation in the August Gwen Challenge.

preface

In the wechat group today, there was a discussion about whether the 28 principles of business model in performance scenario design are applicable. He intercepted the chat in the QQ group.

The question I was asked was whether the 80/20 rule applies to business model design in performance scenarios.

I asked him: In what way does the 80-20 rule apply to performance scenarios? Is there a source? Is there any evidence and data to prove it?

Then there was a bit of a search. But there is no definitive provenance.

The reason for writing this article is that I think a lot of people confuse theory with reality, and simply apply what they think is the right theory in a specific situation without giving details.

The origin of the 80/20 principle

Take a look at the source of the 80/20 rule (from search engines) :

Baledo’s Law (also known as the 80/20 rule) was discovered by Italian economist Roberto Baledo in the late 19th and early 20th centuries. In his opinion, in any group of things, the most important is only a small part of it, about 20%, the other 80%, though the majority, is secondary, so it is called the 80/20 law.

In 1897, The Italian economist Paletto found in his statistical analysis of the wealth and income of all social classes in 19th century Britain that 80% of the social wealth was concentrated in the hands of 20% of the people, while 80% of the people only owned 20% of the social wealth, which is the “80-20 rule”.

In the WiKi, it is explained as follows:

The Pareto principle (also known as the 80/20 rule, the law of the vital few, or the principle of factor sparsity)[1][2] states that, for many events, roughly 80% of the effects come from 20% of the causes.[3] Management consultant Joseph M. Juran suggested the principle and named it after Italian economist Vilfredo Pareto, who noted the 80/20 connection while at the University of Lausanne in 1896, as published in his first paper, “Cours d ‘economie politique”. Essentially, Pareto showed that approximately 80 percent of the land in Italy was owned by 20 percent of the population.

In addition, whether it is Baidu or Google, I have searched the relationship between performance testing and the 80/20 principle, and I have changed several keywords. Basically, all of them are general, that is, there is no data illustration of a case, and there is no specific model derivation process and mathematical support.

In my opinion, Baledo’s Law is just a statistical analysis, not a complex mathematical model.

As an aside, I personally feel that foreigners are very fond of doing this kind of research, and then summarize it. If you have a title and write a paper, you can call it XX law XX theory and leave a name in history.

Business model and 80/20 relationship

With this in mind, I would like to say that in a specific business scenario, the business model in the performance scenario has nothing to do with the 80/20 principle!

Even on a macro level, it’s a stretch, at least so far I don’t see any data supporting it.

Baledo’s law (principle of 80) comes from economics, from the macroeconomic point of view, after the survey sampling statistics, say 80. But in a particular scenario, that doesn’t necessarily apply.

This is especially not applicable to specific application examples such as performance scenario design.

From my years of performance testing and performance analysis projects, most scenarios do not conform to the 80/20 rule. Look at two examples (from 7DGroup and used with permission) :

Example 1: from the following system of some business statistics (day), does not conform to the 80-20 principle (80-20 principle is the power law distribution, and its corresponding is the normal distribution, (do not know what the normal distribution looks like, search by themselves).Example 2: the statistics (days) made by the subway system on the one-card office workers obviously do not conform to the 80-20 principle.One of the more serious guys also found out that there was a book that said the performance scenario was calculated by the rule of 80, and explained that it was to test the capacity of the system. And “Can it be wrong?”

If I write in a book without taking it seriously, I criticize it harshly (though it doesn’t help that I criticize).

How to design the business model

So what’s the right way to design a business model?

Two ways:

  1. For the system with production history data, it must be analyzed from the production history data, and then deduce the business model;
  2. If there is no production history data, first of all, it is based on experienced business personnel to give out, if not, can take the data of the same industry for reference. In other words, a slap on the head is more reliable than the 80/20 rule.

Some people say, we don’t have historical data, no one in the company has done a similar system, and we don’t have access to industry data. Well, I wish you good luck.

In every lecture, I always mention using historical production data to analyze, but I still see many people do not do this even though they have historical data, and it is common to take words out of context.

Take business statistics,

First, business statistics over a long period of time. As shown in the figure below, the data of nearly one month is taken. The longer the data is, the better. The most important thing is to cover enough business scenario days.

Is this already a problem? In fact, this graph is the result of log extraction, second/minute/hour/day summary, percentage calculation and sorting.

Let’s zoom in on the final percentage.

Just look at the last column and take the higher percentage of business. Here we take business 5. These five businesses already account for 94 percent of the total business volume. As a normal day business scenario, this is enough.

In fact, according to the statistics, take out the first five businesses of the day with large business volume (this sentence I secretly changed the concept, but does not affect the rationality of this article, please imagine and balance the picture of a good state of mind). Do the following statistics:Here’s another example:The process of obtaining the above data is as follows:

  • First, pull out all the data for the day with the highest volume of business.
  • Merge on an hourly basis (I’ll make a hole here again, because doing so will reduce the amount of burr data and may miss some special scenarios, so I won’t do perfectionism here, but let’s get the main logic straight).

Let’s also figure out if it fits the 80-20 rule. Let’s do it in percentage terms. For convenience, I’m going to do it in order, just count the first hours, so don’t worry about the order:

By the time I chose 4.66 %, that’s 79.11 %. In terms of time, it is 12 hours, and the service system provides service for a total of 20 hours a day. So the result of this statistic is that 80 percent of business is done 60 percent of the time. (If anyone here asks where 60% comes from, I can vomit blood. Picture)

This is obviously not the 80/20 principle, does it mean the 60/40 principle?

So for a particular business scenario, if the result is 20 or 80 based on statistics, THAT’s fine with me. It is obviously irresponsible to do business model statistics on the basis of 20 or 80 without doing anything.

Now that I’ve done that, I’m going to extend it by hand.

Take the percentage of business in a given hour according to the figure above. The results are as follows:This is where you can set the thread or user ratio in your script.

From the logic of the scenario, the above description is complete enough.

But… The above is really just a business model for a normal daily performance scenario.

This is not enough for a complete system performance scenario design.

summary

We need to do additional statistical analysis to get other performance scenarios, and these need to be evidence-based, polite, and contextual. We will continue to fight against irresponsibility.