Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We have an web server application and web service,which will be hosted in the same server. Customer's IT team has asked us what is the expected no.of traffic on the web server. How do I calculate this?

share|improve this question

migrated from Sep 18 '09 at 11:47

This question came from our site for professional and enthusiast programmers.

Do you want to calculate current traffic or expected traffic? Because those are two completely different things. – Sasha Chedygov Sep 18 '09 at 6:22
I think this may be more suited for server fault as it is a question about server managment not programming. – Jeremy French Sep 18 '09 at 6:22
Don't retag when you can leave a comment and/or flag instead. – random Sep 18 '09 at 6:44
we started to answer so .... – RageZ Sep 18 '09 at 7:12

try this website, to moniter server traffic

share|improve this answer
This dosn't answer the question, it was asking about expected traffic. – Jeremy French Sep 18 '09 at 8:44
he have not clear in the question whether he want to calculate current or expected traffic? so why down vote? – Muhammad Akhtar Sep 18 '09 at 9:16

you can apache/IIS/lighthttpd/etc.. log parsing program to know the traffic. The most well know is awstat.

The log parser will produce several reports with number of hits, page view, traffic,

You can also look into ntop if you need a solution that doesn't involve logs.


share|improve this answer
This dosn't answer the question, it was asking about expected traffic. – Jeremy French Sep 18 '09 at 8:43
yeah sorry just noticed .... – RageZ Sep 18 '09 at 9:27

Calculating expected traffic is not an easy task, the best way would be to have some sort of network traffic monitor installed, setup 10 testing engineers and tell them to browse the site and then collect bandwidth data used by 10 users. Once you have this data, you will get per user usage and then you can determine other parameters like how it will drive users? I mean how do people will know about your site and they will come? by ads? or you already have fixed number of users visiting your site?

share|improve this answer

The key here is that you have to measure the current traffic (per session and/or user) and trend that based on the anticipated number of future sessions and/or users. While it should be easy to measure the current traffic levels, anticipating future growth will be a little trickier and some part of the answer falls on the customer. How many sessions and/or users do they expect to use the site in the next month, 6 months, 12 months?

The customer asking you to estimate what the future traffic level will be is analogous to them asking you when the lightbulb in the broom closet is going to burn out. You can give them an approximation based on the rated life of the bulb and it's current useage, but it's hard to predict what might happen in the future. Will the light bulb be turned on more or less frequently in the next month, 6 months, 12 months?

share|improve this answer

If you want a quick and dirty method...

  1. Disable all services on the server that might be generating/receiving traffic other than what you want to test.
  2. Get a farm of test clients (as suggested by Akash), whether human or script.
  3. Assuming you don't have a Linux-based router in line, get a Linux box somewhere between the clients and server (SPAN/mirror port, dumb hub, whatever).
  4. Watch the traffic. Using a Linux box will allow you to limit the traffic you count based on host, port, etc. as opposed to just looking at the switch counters. You could either use tcpdump or something quick-and-dirty like iftop, bwm-ng, etc. Probably the best would be to use tcpdump and then write a script to read the dump file and only count what you want.
share|improve this answer

You actually have to define traffic in order to answer this question. Do you mean in bandwidth, or in users? Both questions are related, but can answer different questions - it sounds like you're being asked to help with overall capacity planning, which would include both those and potentially resource utilization across the board.

To start with, it may be a question of simply scraping your logs as others have indicated (ie, looking for 'hits' as a metric, or added some basic tracking into your application itself. You can then correlate this with bandwidth used on site, and estimate based off your average number of users what your bandwidth usage would be over time as you added additional traffic/users.

Once you have baseline data for your user base, cross referencing this with metrics like bandwidth, cpu usage and memory are very useful for answering the types of questions you're asking. Munin or cacti, as others have suggested, are excellent tools to begin collecting data and graphing it over time to see trends.

share|improve this answer

How can you calculate expected traffic without a historical trend of traffic on the system? I would implement one or more of the following:

  • Windows Perfmon for Network Bandwidth
  • AWStats or similiar log parsing
  • Cacti SNMP graphing of network interfaces

You can't know the future without studying the past. :)

share|improve this answer

ratio like P/E value compared to similar companies. technically you can profile os or just logfile analyzer or oldschool grep through logfiles are used. depends on technical load or market penetration. first probably serverside measures and latter surely most reliable always validate via external service, don't measure yourself, and compare relative others instead of absolute value.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.