egsemsem
asked on
ISP performance
hello,
I am developing a simple program to test the performance of several ISPs, the test results should be readable by simple Internet user.
I have two methods. Firstly is to Ping some fixed IPs everytime the user connects to the Internet, then I calculate the average Ping time and sort the ISPs according to their average Ping time. The second method is to download a small file (60 kb) everytime the user connects to the Internet, and calculate the average speed. (the average speed in my country doesn't exceed 10 kb\s)
My questions are:
1) In Ping test, which IPs should I choose so that they cover the most important parts of the world?
2) I am also going to divide the day, as you know the Interent traffic varies, what do u think r the intervals that the day should be divided to according to the Internet traffic?
3) Do you think the Download test is good enough to give true info about ISPs performance (to some extend)? And do u suggest that I choose a specified file to download?
I realize that one test isn't enough at all, but I think over an interval of time (2 weeks for example), these results will be meaningfull. The Internet in my country is free (the ISPs take their shares from the phone bill), so the user can use several ISPs per day.
Any good reliable answers or links to good resources are welcome.
I expect there may be several useful answers, so I will give seprate points (in a seprate questions) for other good answers.
thank you,
Osama
I am developing a simple program to test the performance of several ISPs, the test results should be readable by simple Internet user.
I have two methods. Firstly is to Ping some fixed IPs everytime the user connects to the Internet, then I calculate the average Ping time and sort the ISPs according to their average Ping time. The second method is to download a small file (60 kb) everytime the user connects to the Internet, and calculate the average speed. (the average speed in my country doesn't exceed 10 kb\s)
My questions are:
1) In Ping test, which IPs should I choose so that they cover the most important parts of the world?
2) I am also going to divide the day, as you know the Interent traffic varies, what do u think r the intervals that the day should be divided to according to the Internet traffic?
3) Do you think the Download test is good enough to give true info about ISPs performance (to some extend)? And do u suggest that I choose a specified file to download?
I realize that one test isn't enough at all, but I think over an interval of time (2 weeks for example), these results will be meaningfull. The Internet in my country is free (the ISPs take their shares from the phone bill), so the user can use several ISPs per day.
Any good reliable answers or links to good resources are welcome.
I expect there may be several useful answers, so I will give seprate points (in a seprate questions) for other good answers.
thank you,
Osama
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
> You probably ought to use traceroute (instead of ping)
How can I use traceroute to test the ISP performance rather than Ping? Could you give me more details?
> However, I'm sorry to say that my opinion is that your findings will be "representative" of only your own environment
And that's exactly what I want, I don't aim at publishing the test results. I am designing a program to help Internet users find the best and most suitable ISP for their work, it is enough to be "representative" of only the user's own environment.
How can I use traceroute to test the ISP performance rather than Ping? Could you give me more details?
> However, I'm sorry to say that my opinion is that your findings will be "representative" of only your own environment
And that's exactly what I want, I don't aim at publishing the test results. I am designing a program to help Internet users find the best and most suitable ISP for their work, it is enough to be "representative" of only the user's own environment.
There are lots of sites that explain how traceroute works but essentially, the hopcount to the destination is set at one causing the first hop to reply with its address and the RTT. Traceroute then sets the hopcount to 2 in the traceroute packet . . . and so on.
Here's ping:
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
Reply from 10.95.162.38: bytes=32 time=65ms TTL=249
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
And here's traceroute (tracert on Windows machines):
Tracing route to 18.95.162.38 over a maximum of 30 hops
1 3 ms 4 ms 4 ms 10.49.89.253
2 10 ms 11 ms 10 ms 10.49.87.252
3 14 ms 12 ms 11 ms 10.49.139.251
4 54 ms 52 ms 51 ms 10.51.84.182
5 60 ms 62 ms 58 ms 10.95.162.38
Traceroute will provide response times to intermediate stops enroute to the final destination. From this traceroute you can see that the "slow" response time is between 10.49.139.251 and 10.51.84.182. Also, packets travelling on the internet will very often take an asynchronous route so a simple ping may be misleading.
Steve
Here's ping:
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
Reply from 10.95.162.38: bytes=32 time=65ms TTL=249
Reply from 10.95.162.38: bytes=32 time=60ms TTL=249
And here's traceroute (tracert on Windows machines):
Tracing route to 18.95.162.38 over a maximum of 30 hops
1 3 ms 4 ms 4 ms 10.49.89.253
2 10 ms 11 ms 10 ms 10.49.87.252
3 14 ms 12 ms 11 ms 10.49.139.251
4 54 ms 52 ms 51 ms 10.51.84.182
5 60 ms 62 ms 58 ms 10.95.162.38
Traceroute will provide response times to intermediate stops enroute to the final destination. From this traceroute you can see that the "slow" response time is between 10.49.139.251 and 10.51.84.182. Also, packets travelling on the internet will very often take an asynchronous route so a simple ping may be misleading.
Steve
Regarding intervals: I'd perform my tests continuously in order to find out when the peak and off-peak times are. They will, of course, be different throughout the globe.
There's a freeware product called MRTG that you can use to graph your results . . . it will take a little bit of programming effort to format the traceroute data to fit the MRTG model, but it will be worth your time to investigate this. MRTG outputs HTML graphs so that you can display your data on the internet if you like.
Steve
There's a freeware product called MRTG that you can use to graph your results . . . it will take a little bit of programming effort to format the traceroute data to fit the MRTG model, but it will be worth your time to investigate this. MRTG outputs HTML graphs so that you can display your data on the internet if you like.
Steve
You might try traceroute.org. This is a public trace route server site. If you are looking to run ping or traceroutes, this allows you to issue the command from the traceroute server to you (your address should be the same).
You might be able to capture these results and analyze which ISPs the traffic passed through. Hope this is of some assistance.
You might be able to capture these results and analyze which ISPs the traffic passed through. Hope this is of some assistance.
ASKER
Thank you all for your assistance.
stevenlewis please find your points here: https://www.experts-exchange.com/questions/20336806/For-stevenlewis-Q20334197.html
dkpalais please find your points here:
https://www.experts-exchange.com/questions/20336807/For-dkpalais-Q20334197.html
Osama
stevenlewis please find your points here: https://www.experts-exchange.com/questions/20336806/For-stevenlewis-Q20334197.html
dkpalais please find your points here:
https://www.experts-exchange.com/questions/20336807/For-dkpalais-Q20334197.html
Osama
these should give you some ideas
http://computingcentral.msn.com/internet/speedtest.asp
www.dslreports.com/stest
www.bandwidthplace.com/speedtest