3

I have a dedicated server with windows server 2008 and I uploading there a website, I want an estimate number about how many concurrent users can access the website before it crash.

Is that possible?, and how? and which configurations i will use to calculate this?

3
  • I looked at the answers you quoted and had only two concerns...one, apache was asked about and the OP is running Windows which makes me think (since he or she doesn't say) that it's IIS/ASP app, and two, any local testing wouldn't really take into account the affects of your remote users over your connection. It's great that your particular app could server 1,000 users concurrently before dying but once you factor in the pipe connecting your site I would think that practically speaking you'd service a smaller number of people before the site seems to die. Sep 12, 2011 at 18:13
  • Basically I would think the best option (quoted in the question/answers you provided @Dana) are the services that simulate users remotely, as it can test the application and not just your web server component. I'd just be afraid that some of the utilities are akin to defining your server's performance by benchmarking a subsystem alone. Sep 12, 2011 at 18:14
  • @Bart You're correct, those questions don't apply. Comment deleted. Sep 12, 2011 at 18:19

1 Answer 1

2

You would need to find (or hire a service) that simulates users using your website concurrently. You can't do it in general without knowing how much data can be pulled off of it, and that's very specific to your application. How much static content is pulled? Video? Graphics? Generated data?

Then you'd need to figure out how much data can fit through your Internet connection, if your application is doing server-side work whether it'll max your processor or memory, are you using a database? What about your disk storage system? Have you normalized the database?

Most decent hardware today with a decent connection to the Internet should work for most server work, as I doubt most startups are going to hit Google or Youtube or Facebook traffic overnight, so I'd focus more on having an application and server designed to scale up as needed. You can get some idea of the hardware used on Serverfault through the Stack Exchange blog, and they're not really using a lot of heavy duty hardware to serve the sites. But again; it's highly dependent on the application design and the data being pushed to users.

In the end the only "easy" way to tell is to test it. There is no magic formula or program to test it locally.

2
  • do you know a service I can use to do this simulation? Sep 12, 2011 at 18:16
  • I just googled for "web server test sites" and found an aggregator of resources. Probably others if you use terms like stress test web server or some variation thereof. I don't want to recommend a site as I haven't used them. Sep 12, 2011 at 18:24

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