Can xampp handle 1500 connection per hour simultaneously?

[yukija]

Can xampp handle 1500 connection per hour? I mean if 1500 users login to website which running on xampp will xampp work properly?
If not then how many connections? If yes then how many upper limit?

[nigele2]

yukija you need to give more information and even then the answer is likely to be try it and see.

1500 connections per hour will give a number of concurrent users based on how long the sessions last.
The answer might also be impacted by how active the sessions are.
And of course what percentage of available resources each user uses.

Of course even if your server handles 1500 connections it may be that the service offered is so slow your users either disappear due to boredom or suffer timeout issues.

As a first step I would set up a simple script to crudely throw 25 connections in a minute at the server and see what happens. If that works you can then get more sophisticated. There are several free scripting tools on the web that would allow you to script simple usage (e.g. connect, wander a few pages, log off) and will give you some idea of service times.

In clear cut cases the system will simply crash or refuse users at a certain level. Hopefully it will not crash and simply degrade gracefully. With the latter your question becomes "at what throughput (number of sessions and activity) does the service level become unacceptable.

In the last few weeks I have tested 20000 concurrent users on an apache based web service with very high throughput - all things are possible but as a professional performance tester I have to say the only answer is "test it".

Hope that helps. Happy to discuss.

[nigele2]

yukija thinking about it I might give it a try for you. I'm tied up testing a university system in Scotland for the next 10 days but after that I might get a chance.

Have you examples of what the user does? If you make it read only that is much easier (and all I could do for free). I'll need to check I have sufficient upload capacity. Also wouldn't want to handle https (although it can be done it adds to the burden).

Can you send me a url and test user so that I can see the application?

And also I need someone to answer my thread further up about why I have null variables in my php. Anyone?

Hey this ZAMPP stuff might turn out to be fun

[yukija]

Thanks for your reply. I am really talking about XAMPP windows based apache server(not LAMP).

Actually the users are only browsing the sites, sometime create linking since its a Drupal CMS based website so few (may be 50 among 1500) users will have permission to modify contents. And there are 200 video contents (which can be viwed by JWplayer) may be will grow 1000 in future and they are watching not so frequently. May be once a week per user.

Right now 100 users are using XAMPP with drupal and working fine so far (tested and happy with performance). But we are planning to big move i.e. 1500 users. Thats why I am worried. You may ask me why not using LAMP but problem is I dont really have time to configure LAMP and dont know well. So I would rather like to use xampp based apache windows version. Its an intranet based website. So I cant give you the url.

Is there any tools which can test the webserver load (same as 1500 users load) so I can figure out how the site will work when 1500 users login and using xampp based DRUPAL CMS website. Appreciate if you could help me.

[nigele2]

Yukija there are hundreds if not thousands of tools out there ranging from free to quality services including trouble shooting for thousands. As with all things software you are going to have to find some time and effort to get the best out of whatever and I know - "I haven't got any time!".

Many of the utilities will allow you to wander around the site firing urls but clearly you need scripts which will reflect real user behaviour and actually log on and download. That will negate most of the free offerings.

One suggestion I would offer is that if you acquire a tool (or service) make sure that error checking is incorporated into the test drivers and/or process. All to often when I do test audits I see published response times but sometimes 90% are the service times of failures - and most systems return failures much quicker than valid results.

I cannot offer our own tools as they are very much tools (as opposed to a testing package) and as such take considerable learning time.

Perhaps someone passing may have some suggestions but good luck with whatever approach you take.

[yukija]

Appreciate if you post few links of testing tools since I am novice and dont know much...thanks.

[nigele2]

Yukija

OK but can you tell me if you are using AJAX (especially any a-sync).
Are you using https?
I assume you are using html and javascript as the client technology.

Cheers Nigel

[yukija]

OK but can you tell me if you are using AJAX (especially any a-sync).
ANS: YES

Are you using https?
ANS: NO, BUT HAVE PLAN IN FUTURE

I assume you are using html and javascript as the client technology.
ANS: YES

THANKS FOR YOUR TIME.

[nigele2]

Yukija

Sorry for the delay in getting back to you but I am currently tied up testing a university student administration system and it must be finished by this coming Wednesday.

I have however had a look at a few tools but cannot find anything free that will do more than fire urls at your site and that will not help as you have:

users logging on and thus sessions (or whatever equivalent) to handle
AJAX asyncs which need to be valid in terms of data (I guess that depends on what your system does with poorly formed AJAX requests?)

However your question was whether the system architecture could support 1500 concurrent users. To answer that the most important thing is to log them on. What they do matters less as long as they show the correct level of activity.

Even so without people who have played with scripted network traces before I think it will take considerable effort to get even simple tools working in such a way as to maintain sessions. Configuring a more robust web server might in the end be the better bet.

But I will look further and get back to you. To give you some idea (it is just to give an idea as no one is or should pay big bucks to test XAMPP *) a typical test exercise of this type would cost £12000 including test software, expertise and trouble shooting - and take about 10 days).

* I do not mean disrespect to ZAMPP but as I understand it it is offered to support developers and not as a robust production system supporting 1500 concurrent users - but always happy to be proved wrong).

I'll get back with any possible solutions later in the week.

[yukija]

waiting

[nigele2]

Yukija OK had a chance to look around and if I were you I would have a look at: Tsung, curl-loader, Siege and Apache JMeter. All easily found via goggle. (Add 'load test' to siege unless you are into military history).

To say where to begin is impossible for me to say without seeing the application but the important thing is going to be : user authentication and maintenance of sessions.

Would be great to here how you get on. good luck

ps For anyone thinking of trying such tools I would recommend: Use the tools to create a load. In parallel log on real users to appreciate the 'look and feel'. There can be considerable differences between network responses and actual end user experience.

[yukija]

thanks a lot will feedback to you soon.

[nigele2]

yukija how are you getting on?

I have just been working with a University and to get some basic concurrency data they used Jmeter with some success. It was crude and dirty but when we did serious testing the indications they had were not a bad guide.

Anyway Happy New Year.

[Izzy]

Maybe some performance tuning coupled with any benchmark testing would be of assistance as this post gives some useful reading:
viewtopic.php?p=154539#p154539