I have a (simple?) question that I hope someone will find interesting enough to help me out with.

A web site has a given number of subscribers who generate a certain amount of traffic on the web server. Hits (web requests) are distributed throughout the day and result in a peak load of x hits/second.

Question: should the number of users increase (let's say double) and follow the same usage pattern, how would the peak load (hits/second) be affected?

I strongly doubt that the answer is 2x hits/second and realise that more information might be needed, to start from the actual distribution of the hits and the average total hits per day.

Anyone intrigued and kind enough to lend a hand?

Thanks, Paolo

  • $\begingroup$ You'd need to define the peak load. If you model the hits as instantaneous events, there's no such thing as an instantaneous peak load. Either you need to look at the number of hits in a certain time interval, or the requests need to take a finite time to be served -- the result would be the same. $\endgroup$
    – joriki
    May 17, 2013 at 17:21
  • 1
    $\begingroup$ That's not what 'distribution-theory' means! Easy mistake to make, I know, but do read the category descriptions when tagging. $\endgroup$ May 17, 2013 at 17:45
  • $\begingroup$ @joriki - Peak Load = number of instantaneous events that occur within a second. $\endgroup$
    – Paolo
    May 20, 2013 at 9:40

1 Answer 1


It does indeed depend on the distribution and number of hits.

For example, suppose everyone always logged on at the same moment. Then the peak doubles.

Now suppose that the distribution is completely random throughout the day, and that you go from 1 user to 2 users. The chance that they log on at the same time is 1/86400. Therefore the peak will almost certainly stay at 1.

Basically, if the distribution is 'filled out' by the current usage, it will go up by 2, if it's sparse, it will stay the same. In between, it will vary, but on average increase by a factor between 1 and 2.

  • $\begingroup$ Thanks. Indeed my question is how to practically calculate that factor (well, its pdf). $\endgroup$
    – Paolo
    May 20, 2013 at 9:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .