Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I've been using smokeping to monitor internet latency (for early problem detection) at for several months. On Friday the average latency went from 12ms to 55ms and has been at the 55ms plateau for 3 days. Our WAN uses the same provider for all internet connections so we're curious as to whether they have a problem or if just decided to choke the pipe down.

Our ISP claims the problem is at, but I am still suspicious.

I notice I get the same (55ms) latency to, but I wasn't graphing that IP so I don't have any reference data to prove a degradation in service.

Any recommendations on how to pursue this anomaly would be helpful.

PS: Flow shows no excessive traffic on our routers that would point to a behavior-driven drop in service.

share|improve this question is an Anycast address, there's 31 sites currently serving it (Google doesn't publish how many servers there are at each site, might be one or many). Changes in route weights could easily make one route "cheaper" than another changing latency. – Chris S Aug 22 '11 at 21:35

When you see latency change like that, the cause 99% of the time is internet routing changes. Unfortunately unless you're capturing periodic traceroutes, there's nothing you can do to see what the route used to be.

I'd highly doubt that is doing anything directly that caused that jump in latency.

share|improve this answer is Google's public DNS service.

They achieve geographic proximity through BGP anycast, meaning that the physical server that you hit on one day, and the path that you take to get to it, can change drastically depending on where Google's advertising it from and where the shortest path is to that location.

It sounds like your traffic changed from going to a nearby server to a further away one - Google may have taken down the near one, or your ISP's BGP process may see another advertisement of as closer in terms of routing hops to reach the destination.

That's neither here nor there; the point is that is not a good address to use to gauge your internet link's performance, and fluctuations such as this should be expected if you do.

share|improve this answer
+1 for giving me a chance to go and read up on BGP anycast info! – EightBitTony Aug 22 '11 at 22:24
All great responses. Thanks for your help! – josh Aug 24 '11 at 16:50

traceroute will show you how long each hop takes, and you can work out where the biggest contributor to your 55ms rountrip time sits. As ErikA says, without previous data, it won't help much but at least you'll have some idea.

share|improve this answer
Some ISPs give these packets the lowest form of priority. This means that at times you may get very bad results. I'd recommend continously monitor the values over a larger period of time and then compare the results, in median form. – artifex Aug 22 '11 at 20:53

Your Answer


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.