[Outages-discussion] [outages] PINCH Re: Slack Outage?

Scott Whyte swhyte at gmail.com
Mon May 21 19:04:32 EDT 2018



On 5/21/18 2:43 PM, Jay R. Ashworth wrote:
> ----- Original Message -----
>> From: "Scott Whyte" <swhyte at gmail.com>
>>>>> Slack is only barely on-topic for the list anyway.
>>>> Explain that a bit more please?  From the about page
>>>> "outages-reporting that would apply to failures of major communications
>>>> infrastructure components having significant traffic-carrying capacity,"
>>> Slack being an end-user service, it's on-topicness falls under "eyeball services
>>> that our audience is likely to get lots of reports about" -- with a couple
>>> extra points because lots of the people on the list probably use it themselves.
>> Ok so you are differentiating between transit and end traffic, despite
>> the fact that several "eyeball" networks source/sink way more packets
>> than any ISP nowadays.  I can see the value of this differentiation (if
>> I squint), maybe we need a eyeball-outages to better separate the two?
>> Or is there one I'm not aware of?  I'm tired of hearing about single
>> circuit outages, but I'm super tired of all the whinging.
> Not eyeball network -- eyeball *services*.
>
> A transport outage inside Spectrum is clearly on-topic.
>
> A transport outage at 111 8th St is also on-topic, even though it's in a
> single building.
>
> An infrastructure outage at Amazon East is on topic.
>
> An outage of a single service hosted by Amazon (or whomever) depends a lot
> on what that service is -- how many trouble calls is it likely to generate,
> and how important to the maintanence of the network that service is.
>
> Flickr, for example, wins on the first point; Slack on the second.
>
> That's the way I've been looking at -- and handling -- it, to date.
>
> Does that seem reasonable?
Man after an hour of kicking this around I feel your pain.  We'd like to 
discern between underlay, overlay, and service, but wtf do those terms 
even mean in the modern network where any layer can and is transported 
over another layer?

For me, pulling what are "clearly" services :) off into another list 
might be the most helpful and least disruptive.  In this case I'd 
suggest anything that is a service would be fair game on the new list, 
to lighten the moderation load, similar to how on this list anything 
"clearly" underlay is fair game as well.

Thoughts?

>
> Cheers,
> -- jra



More information about the Outages-discussion mailing list