Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

(There's a lot of new info in the updates.)
I've been getting very high pings when dealing with some servers, and I can't figure out the cause. For the question, I'm using a specific IP, but I have verified this issue with several IPs starting in 189.1.

If I open the windows command prompt and type ping -t 189.1.164.176, I get latencies of 170ms on average. The problem is: the location for that IP (São Paulo) is only a couple of states away from me, and the latency should be less than 50ms on average. I need to know what's causing this, so here's what I've investigated so far:

  1. A friend of mine lives in the same city as me (less than a mile away), and hires the same internet service provider as me. When he pings that IP, he gets an average latency of 40ms. That indicates it's not an issue with my ISP, and not a geographical issue either.
  2. All computers in my home network suffer from this issue. I tried it on both wired and wireless computers. That indicates it's not an issue with my PC.
  3. Other people all accross the country (as well as my friend), don't suffer from this. That indicates that it's not a problem on the server's side (though I'm not sure).
  4. This issue is constant and permanent. It's very stable too, always at levels around 170ms. That indicates it's not an issue of the server being overloaded or anything.
  5. I thought it might be an issue with my router, but I only get this problem with a small set of IPs, any other IPs I tried have great latency. Would it make sense that my router is discriminating one select set of IPs? I haven't configured any IP related stuff in it (only port forwarding).

And that's pretty much it. I'm out of ideas. I used "indicates" in every sentence, because I'm no expert and I could be wrong. What could be the cause of this issue?

It's driving me nuts!

Update:
I ran winMTR here at my PC, trying several different IPs that are problematic. Analysing the average ping from each step, here's what I got. They all get the high ping increase at the 6th step.

  • Ping is 30 on the 5th step. The host name is c90601e2.peer.uol.spo.virtua.com.br (virtua is my ISP, the rest is cryptic to me).
  • On the 6th step it jumps to 150. There is no hostname, the IP is 200.221.136.124.
  • There are 3 more steps after this one, but they don't show anything interesting.

All problematic IPs showed this behhavior. The problem is clearly here. A search on webipaddress.net reveals that IP belongs to UOL (another ISP from here).

**SECOND UPDATE**

Here's where it gets weird. I asked my friend to perform the exact same test on his machine. And his route is almost identical to mine! He goes through the exact same hosts in steps 5,6,7,8, and 9, but there's no high ping. Just to be clear: his route goes through server c90601e2.peer.uol.spo.virtua.com.br and also 200.221.136.124 but doesn't suffer from high ping increases in that step (like mine does).

What could possibly cause this behavior? Why are MY packets getting slowed by that specific host, but my friend's packets aren't? What could distinguish between them?
I'm already contacting tech support, but I don't really expect them to solve anything.

Thanks to those who answered, it really helped nail down the issue. I only wish I knew why it's happening.

share|improve this question
    
Email them your output instead. I'm glad someone told you (and me)about MTR. It's nice to get taken to school on someone elses question... :) –  Everett Nov 20 '10 at 6:47
    
Yeah, that's probably easier. I'll check if they have a contact e-mail for tech support. –  Malabarba Nov 20 '10 at 7:25

2 Answers 2

up vote 3 down vote accepted

A friend of mine lives in the same city as me (less than a mile away), and hires the same internet service provider as me. When he pings that IP, he gets an average latency of 40ms. That indicates it's not an issue with my ISP, and not a geographical issue either

This doesn't matter, he could be routed through a different CO, or he could be on a different ATM or Trunk that has lower latency or some routing issues. If you have XO, you almost certainly have routing issues.

All computers in my home network suffer from this issue. I tried it on both wired and wireless computers. That indicates it's not an issue with my PC.

This is pretty standard as all the computers in your house will take the same route.

Like already suggested, run a test, traceroute is not a good idea, it won't tell you anything because you know the route completes and it really only gives you a brief snapshot, what you need is MTR or pathping to tell you the latency at each hop over a period of time. This will produce much more accurate results. If your ISP is peering with the network that is returning high ping times or it is on their network, they can redirect you (although it might take stubbornness on your part to get them to do it). If it is off-net for them, you're out of luck.

share|improve this answer
    
I beg to differ. Traceroute IS a good idea BECAUSE the route completes. This could show you 1) routes that change. I've seen this problem before with routing that was unstable (the route kept changing). 2) routes that loop. I've seen routes loop between providers a few times before being handed outside the loop. 3) Traceroute can show you latencies. If they vary wildly, this will be noticeable. All of this would be useful information in diagnosing where the failure is, no? –  Everett Nov 20 '10 at 3:44
1  
pathping and MTR will show the same thing, and more, why go for less information? Traceroute shows a snapshot, one instance that may or may not be high, when the other two will give you a more standardized idea of how the latency fluctuates. I do troubleshooting for an ISP and we NEVER use traceroutes to troubleshoot routing issues as they are pretty much useless compared to MTR. –  MaQleod Nov 20 '10 at 3:54
1  
I'd like to check with MaQleod on this, but I wonder if your friend has a lower MTU setting than you do, and maybe, just maybe you are experiencing a small amount of fragmentation. Any thoughts @MaQleod? –  Everett Nov 20 '10 at 20:18
1  
You can usually set the MTU size in most routers in the advanced settings. Most of the time, you have a specific reason to change the MTU and most home routers use the same default settings. One thing you can do to test if different MTU is to change your ping settings: ping google.com -n 100 -l 800 (on windows) or ping google.com -c 100 -s 800 (on linux/unix). I have a feeling that there will be no change in latency, but it is possible that packet size has an effect on the router that is causing the problems on the path. –  MaQleod Nov 20 '10 at 22:35
1  
Looking at your results, if the increase is happening at a certain host all the time and that host is a part of your ISPs network, then you need to contact them. That particular device is a router and it might be a card issue. All of these types of routers are large rack-mounted devices that contain a number of cards, each card has a number of ports. If you are on a card that is experiencing problems, only those on that card will see the same issues. It is possible it is even just your port. This is something that only your ISP will be able to diagnose. –  MaQleod Nov 20 '10 at 22:39

I have tried WinMTR from Europe to 189.1.164.176.

While the route taken is of course substantially different than yours with 17 steps, I was struck by one particularity, where the package was transferred from 200.221.30.94 to 200.221.30.94, and with packets lost in addition.

Using whois, 200.221.30.94 is resolved to "Universo Online S.A." which I guess is your ISP.
(Correct me on this, since my conclusion is based on this fact.)

My conclusion is that your ISP has lousy internal routing tables that shuttle packets needlessly between its servers, and in addition is losing packets. If that is the case, you should be thankful that your Internet connection is at all functional.

I would suggest that you make more tests to establish some solid evidence before getting in touch with their Support. If nothing comes out of it, and if you have the choice, another and more competent ISP might be a good idea.

share|improve this answer
    
Universo Online (UOL) is not my ISP, but it's another ISP from here. I'm getting my ping spike from them as well, so it might be a hardware issue from them. –  Malabarba Nov 20 '10 at 22:55

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.