Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

First of all, I hope my post are is posted in the right section . Below are something i got confused hope you one can help.

My business partner 's IP server are in: Sanjose, CA , and I am looking for a datacenter which give the least latency to that IP. I found 2 . 1st datcenter is in San Francisco, CA . 2nd data center is in Newyork.

Then I do the ping from each datacenter to the my partner's IP.

1st datcenter's result: 75ms

2nd datacenter's result: 2ms

I 've done multiple time, 2nd datacenter always give better result. Now my question is:

Isn't the 1st datacenter suppose to give better result, since their location are closer ? How come the result produce different, what is the accurate way to check latency ?

share|improve this question
8  
Something's not quite right. San Jose to NY is about 4000km, the speed of light requires that the round trip time is no faster than 26ms so either someones suspended the fundamental laws of the universe or either DC 2 is not really in NY or the target is not in SJ. It seems much more likely to me that the target ip address you are testing is physically close to New York and not in San Jose. –  Helvick Jan 7 '11 at 20:49
    
@Helvick lol, good job at doing the math that none of us wanted to do :) –  Michael Lowman Jan 7 '11 at 21:21

4 Answers 4

To check latency over time, I recommend Matt's Traceroute: http://www.bitwizard.nl/mtr/

That being said, geography is not always the determining factor for latency. Latency is often caused by the number of network hops it takes to reach your target machine, but there is a large array of factors that can come into play.

share|improve this answer
    
+1 for mtr being a great way to figure this out. –  Phil Hollenback Jan 8 '11 at 1:10

Latency is dependent upon the quality of the connection between two points. If the connection goes through a lot of processing through routers and switches, especially heavily used or underpowered ones, the latency will increase. Distance is not the sole consideration.

share|improve this answer

It's not just about distance and the speed of light: the physical implementation of each interface along the route might have a bad influence of the trip time.

It looks like your route to NewYork jumps on a nice backbone straight away and your physical location in New York has a very good connection to that backbone.

Also, for reliability's sake, you might want to think about the St. Andreas fault line.

share|improve this answer

While the first datacenter may be physically/geographically closer, that doesn't mean that the path is shorter. Run a tracert for/from each location.

My datacenter is 12 miles from my office. My ISP's connection from my office to our T3 provider in our datacenter routes my traffic from Cleveland to Chicago and back to Cleveland.

Also, how is it that you're running a test FROM each datacenter?

share|improve this answer

Your Answer

 
discard

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.