PDA

View Full Version : We love InterNAP, our data center and network provider. Here's why


FH-Dave
05-01-2003, 07:11 PM
Here's an example of what Internap can do for us using their intelligent routing technology:

We submitted a message to Internap Network Operation Center about ping problems experienced by one of our customers in Turkey. Here is how it was handled (in less than 30 minutes)

It looks like the issue is due to the return-path chosen from our BSN PNAP
(starting at the 208.172.63.78 IP in your trace). It appears that there
are some significant packetloss and latency issues with AS 9121 (TURK
TELEKOMUNIKASYON A.S.) and their connectivity to the US. I'm seeing
performance issues reaching their network over Verio, C&W, and Sprint, as
shown below, so I have moved our return-path traffic over to UUNET,
improving things significantly.


| Backbone Average StdDev NumRecv NumSent %Loss |
| <verio, 2914> 572.96 7.57 91 115 21.0 |
| <cw, 3561> 395.71 6.5 94 115 19.0 |
| <uunet, 701> 396.2 37.86 115 115 0.0 |
| <sprint, 1239> 580.95 8.43 90 115 22.0 |
| <gblx, 3549> 396.82 38.27 115 115 0.0 |
| <genuity, 1> 396.55 37.88 115 115 0.0 |


Unfortunately, our performance measurement system has been unable to
collect performance data for this network (most likely due to ICMP
blocking), so our routing optimization system has been unable to properly
optimize this netblock. I have opened ticket number 118229 to track this
issue. If you see persistent performance issues to this netblock, and if
we can obtain a highly available IP address within the 213.74.0.0/16
netblock that accepts ICMP pings, we may be able to have our performance
measurement system gather performance data to use in optimizing
connectivity to that network. Alternately, if we see the same performance
issues over the same networks repeatedly, we may be able to limit the
system's available choices to just those networks that don't show a
problem.

The changes I've made to the routing of 213.74.0.0/16 from BSN will remain
in effect until between 02:00 and 03:00 tomorrow morning, when our routing
optimization system will calculate a new best path using whatever
information is available to it at the time. If the problem returns
tomorrow, please let us know and we'll look into implementing a more
permanent solution.


How many providers can do this for you? Fluid Hosting is very satisfied to be able to extend this level of sophistication and performance to its customers.