Issue 17: DNS Timing and Smokeping distributed monitoring

November 16, 2007

Users of Smokeping know that Dynamic Network Services Incorporated recently sponsored a number of enhancements for the 2.2 release. The feature we are most proud of is the remote monitoring functionality. For the first time in smokeping (and in almost any application), you can see latency from different locations.

DNS can primarily be measured by latency: the average speed to answer a DNS query. The goal of any network is to have the lowest average and limited distribution in response times. Striving for that level of performance on a global scale means a consistently fast user experience so that the end-users of our customers can access Internet resources faster. We do that by strategically choosing our locations and bandwidth carriers.

We have seen some users write to us regarding DNS Timing reports or Zonewatch reports. They rely on those for quantifiable and accurate numbers for their DNS performance. The truth is that latency is a hard to measure. First, you need to consider your mean audience. The current timing tests are conducted from what are considered core parts of the network and generally on bandwidth much different from where the eyeballs are coming from. It’s much further out in the network. Second, if I am measuring from two locations, say NYC and the middle of the Sahara, I need to factor in population which those reports do not do.

Dynamic Network Services is constantly measuring, evaluating, and improving mean resolution time. It’s sad that there is no one key, uniform metric that you the user can rely on for an accurate measurement of DNS performance. With tools like this, we do take a step closer. We take these type of tools seriously (perhaps too seriously :)) and hope to improve on the understanding of this topic.

Need DNS that is particular quick? Check out Dynect – anycast, superfast DNS.

DynDNS Network Monitoring Available Monthly →
← Linux/Unix Clients That Work - ddclient and inadyn