Akamai shutdown, network interruptions

Over the past couple of days, we’ve had our website offline, remote management access broken, and other strange things happen.  Some of it was related to Akamai, who was sending lots of traffic not to MBIX members, but out to general transit, in volumes much higher than we were prepared to deal with on short notice. As a result, circuits were shutdown, and access was cutoff.

We are back to being fully operational now, with IPv4 transit from Les.net and Rainy Day, as well as IPv6 transit from HE.net.

For now, we have disabled transit for the Akamai cache, until we can have a technical discussion with Akamai about how their node should be expected to behave. When we have good answers, we shall enable their connection again: we expect more modest traffic in future.

Akamai CDN node engages thrusters

Akamai, AS20940, has turned on their caching node at MBIX. We can tell they’ve turned on because of the large amounts of traffic now flowing through MBIX.  Having Akamai locally in Winnipeg should be a good benefit, allowing members to transfer data from this large content distribution network directly, lowering cost and improving performance.  Akamai has lots of capacity, and we welcome new members to join and see how their traffic patterns change for the better. 😉

Check out the stats page to see the data flow.

Statistics now available

We now have public graphs showing aggregate traffic at MBIX.

These will be permanently available on our statistics page.

Past 24 hours:

Graph of traffic over past 24 hours

Past 7 days:

Graph of traffic over past 7 days

AS23523 peers at MBIX

Rainy Day, AS23523, has lit up a peering session with the route servers on IPv4. Routes are now visible, and traffic is beginning to flow. Our prior announcement of Rainy Day’s connection was a bit premature–they were physically connected but had not yet BGP.