Re: "Weird" Traffic about 10 hours ago

[Deepak Ravi]

 Based on the data we’ve, it looks like there was a noticeable spike in
packet loss within the AWS's Ashburn and Ohio datacenter around 17:50-18:10
Central Time. It is possible that this had something to do with the drops
and dips in traffic you listed. If you’d like some additional data, feel
free to email me off list.

Disclaimer : I work for ThousandEyes.

-Deepak

On Tue, May 1, 2018 at 8:28 AM, Mike Hammett  wrote:

> This is going to be extremely scientific. 
>
> Did anyone else see "weird" stuff going on about 10 hours ago, about 6 PM
> Central on April 30th? I'm based in the Chicago area and have a large
> client on the east coast. I saw about a 25% drop in traffic for a half hour
> to an hour.
> Another Midwestern ISP also saw about 25% drops on two different upstream
> connections.
> A friend of mine runs an ISP in Virginia. He reported a dip in traffic
> (though didn't report how much of a drop), but also pings and IPSEC\L2TP
> worked, but couldn't SSH or do other activities.
> That friend reported another ISP in Virginia had problems at that same
> time.
> An ISP in Cyprus reported issues at that time. I'm looking to firm up what
> kind of issues and verify the time.
>
>
> Yet other ISPs report no problems at all. Smooth ramps on traffic graphs
> as one would expect at the beginning of prime time.
>
>
> I thought maybe "something" happened in Ashburn (fiber cut, DWDM card
> failure, etc.) as my client has a wave from the east coast to me in
> Chicago, but then the Midwestern ISP shouldn't have any dependency on
> Ashburn, given Chicago and Dallas.
> Then there's the guy in Cyprus, which shouldn't have any bearing on
> anything that happens over here in the States.
>
> I thought maybe it was an epic failure at one of the CDNs or other content
> networks, but then that wouldn't impact SSH or other management activities.
>
>
> Anyone else have any other data to help figure out what caused this?
>
>
>
>
> -
> Mike Hammett
> Intelligent Computing Solutions
> http://www.ics-il.com
>
> Midwest-IX
> http://www.midwest-ix.com
>



-- 
*~Deepak*

Re: "Weird" Traffic about 10 hours ago

[Aaron C. de Bruyn via NANOG]

My extremely un-scientific reply:

I make a lot of connections from Washington State to Virginia every day.

Around 5 PM PDT yesterday, I got booted and had trouble re-connecting for
about 10 minutes.  I figured it was just me, but then a handful of sites
wouldn't load for me while others had no trouble.  I started to do some
traceroutes and they all succeeded, then I noticed my connections to
Virginia were restored.

I don't think it lasted for more than 10 minutes whatever it was.

-A

On Tue, May 1, 2018 at 6:31 AM Mike Hammett  wrote:

> This is going to be extremely scientific. 
>
> Did anyone else see "weird" stuff going on about 10 hours ago, about 6 PM
> Central on April 30th? I'm based in the Chicago area and have a large
> client on the east coast. I saw about a 25% drop in traffic for a half hour
> to an hour.
> Another Midwestern ISP also saw about 25% drops on two different upstream
> connections.
> A friend of mine runs an ISP in Virginia. He reported a dip in traffic
> (though didn't report how much of a drop), but also pings and IPSEC\L2TP
> worked, but couldn't SSH or do other activities.
> That friend reported another ISP in Virginia had problems at that same
> time.
> An ISP in Cyprus reported issues at that time. I'm looking to firm up what
> kind of issues and verify the time.
>
>
> Yet other ISPs report no problems at all. Smooth ramps on traffic graphs
> as one would expect at the beginning of prime time.
>
>
> I thought maybe "something" happened in Ashburn (fiber cut, DWDM card
> failure, etc.) as my client has a wave from the east coast to me in
> Chicago, but then the Midwestern ISP shouldn't have any dependency on
> Ashburn, given Chicago and Dallas.
> Then there's the guy in Cyprus, which shouldn't have any bearing on
> anything that happens over here in the States.
>
> I thought maybe it was an epic failure at one of the CDNs or other content
> networks, but then that wouldn't impact SSH or other management activities.
>
>
> Anyone else have any other data to help figure out what caused this?
>
>
>
>
> -
> Mike Hammett
> Intelligent Computing Solutions
> http://www.ics-il.com
>
> Midwest-IX
> http://www.midwest-ix.com
>