Thanks:  0
Likes:  0
Dislikes:  0

# Thread: Distribution of users for simulate

1. ## Distribution of users for simulate

Hello

I have an information about the number of user that visit a web page in a Day, 13500 users. But I don't have the information by hour.
How can I simulate the normal concurrency of this web site by unit of time?
I am doing a distribution of Poisson, where I Use as mean 250 users by hour (13500 Users/12 hours), with this number I calculation the probability of visits by hour, for this case the probability that visits will be more than 314 users by hour is 0,00%. Then I will simulate this number of users by hour. is it correct?

Can you help me with this?

Thanks,

2. ## Re: Distribution of users for simulate

It is a guess, and without more granular data, it will never get any better. I also think it is rather low, and depending on what your site does, 12 hours is an awfully short day. Can you tell us more about the site?

3. ## Re: Distribution of users for simulate

Well lot depends on the type of the site and business the users conduct.you can determine peak by analysing the business requirement.Also note that number of user visiting the site might be 13500, but it need not mean that all are doing some jobs.Also i disagree with your calcuation of 250 per hour.There is lot of difference between concurrent and simultaneous connection.so i would suggest you to go through requirement, interact with business analyst and come down to figure out the number of users.

4. ## Re: Distribution of users for simulate

13500 user by day is for the home page, because I have others numbers of users by day for some pages in the site, for example, 5600 users by day for another transaction(page of search, without entrance datas, only selection of links).
I want simulate the load of Home page first, and is true that the 12 hours is a short day but this is the amount of hours that the users of application estimated use it.
For this I need to know what is the best method for calculate the amount of users that visit by unit of time.

5. ## Re: Distribution of users for simulate

Your method could be correct, or it could be completely wrong. Knowing only "13500 user per day" doesn't help much in determining what the expected concurrency should be.

For some sites, the majority of users might hit the home page close to the same time each day, or within just a few hours.

For example, if a site is somewhat local, it's possible that everyone would view it as they arrive at work, then not at all for the rest of the day.

It's even possible (though unlikely) that the majority of the 13500 users hit the home page at the same time.

How long do you expect these users to stay? Is it possible that you would have 13500 concurrent users at some point?

6. ## Re: Distribution of users for simulate

The time that I expect that these users stay in the aplication is around to 30 secs, and is possible reached 13500 concurrent users in the same point, because there isn't restriction. But I think that if I'll simulate this amount of user in the same time, I will be doing a stress test at the application, I know that this information is important too, but by now I need to know how is the behavior of the application under normal load, but I dont have more detailed information

Thanks.

7. ## Re: Distribution of users for simulate

A review of your web server log files typically will review actual usage and answer the when, where and how much.
Right now you simply have too little information to come up with a good user load model.

8. ## Re: Distribution of users for simulate

I will check if there are log files, if not, I am going to do the simulation increasing the number of user that visit the aplication until reach the limit of users by hour that they can enter because the aplication not work or when response time has an exponential growth.

9. ## Re: Distribution of users for simulate

[ QUOTE ]
I am going to do the simulation increasing the number of user that visit the aplication until reach the limit of users by hour that they can enter because the aplication not work or when response time has an exponential growth.

[/ QUOTE ]

There are 2 things which i feel you need to confirm, are testing for hits/sec so as determine the actual load on the system, if yes than you can achieve the same with less users and proper setting,
if you are testing for actual users that is by numbers than you need make sure that you are giving realistic screnario in the test.

Also note that response time depends on both these factors.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•
Search Engine Optimisation provided by DragonByte SEO v2.0.36 (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.