Should Netflix and Hulu give you emergency alerts?
Sean Donelan
sean at donelan.com
Sat Mar 9 20:03:02 UTC 2019
On Sat, 9 Mar 2019, Brandon Martin wrote:
> Any reason the ISP has to be directly involved in this? The relevant
> government organization originating the alert could easily have a service to
> make that information available to the public via some standard API (maybe
> they do)?
ISPs with Akamai servers on their network already have this. Akamai
supplies FEMA IPAWS messages anywhere on its CDN.
Smart Speakers, Smart TVs, streaming devices could query the nearest
Akamai CDN server on the ISP network.
> Does it have to be push and application-agnostic? Maybe that's (gasp) a
> reasonable application for Internet multicast. Operators could help, here,
> by making sure that particular application of Internet multicast actually
> works even if other applications don't, and governments originating alerts
> could help by making that straightforward.
I've proposed using multicast in the past. Canada uses IP Multicast over
satellite for its system. Emergency alerts are usually supplied at
no-cost to subscribers by the supplier. That is, cellular carriers do
not charge for SMS fees, cable TV companies do not charge for the
emergency alert channel, etc. It may be part of the overhead cost, but not
a separate subscriber charge.
Although FEMA IPAWS alerts are digitally signed, ISPs have never figure
out a good way to secure IP multicast and prevent IP multicast DDOS. ISPs
also haven't figured otu a good way to charge money for IP multicast. But
since emergency alerts would be a non-chargable service, that's less of an
issue for emergency alerts.
But if ISPs did, they could transmit the FEMA IPAWS alert stream via
multicast. Managed IPTV suppliers, i.e. AT&T U-Verse, use IP multicast for
their emergency alert service.
> Is it sufficient for the streaming services to simply include this
> information in their streams? Heck, they could just include all of them and
> let the device that's accessing the stream figure out which ones are
> relevant. After all, it's the streaming service that knows the user is
> consuming content suitable for inclusion of emergency alerts. The network
> operator rarely knows this directly (though we're pretty good at inferring
> it).
Requiring each streaming service to duplicate the emergency alert stream
seems to be a lot more overhead than ISPs supplying one stream with only
emergency alerts. This seems more like ISPs trying to cost shift to
streaming suppliers, or vice versa, streaming services trying to cost
shift to ISPs.
ISPs already supply a DNS lookup service as part of their Internet
service. ISPs could supply an Emergency Alert service as part of their
basic Internet service. Of course, some ISPs suck at even DNS, so there
are third-part DNS services.
Unlike analog channels, Internet Protocol is multiplexed. IP can handle
data from different sources over the same interfaces.
> Do they? They know where the account is geographically located, but they
> don't necessarily know that the device consuming the media is located at the
> account address.
This was the huge debate with 9-1-1 VOIP emergency phone service. Rather
than repeating the geolocation arguments, just incorporate the 9-1-1 VOIP
geolocation debates by reference.
> Again, operators could help here by providing some sort of service to say
> "Where is my account located?", but many consumers of streaming media have
> far more accurate information based on mobile network geolocation
> information, Wi-Fi mapping, or outright GPS.
I agree, when available.
If a streaming device can figure out how to black-out sports events or
geo-target advertising in specific geographic areas, it should be able to
figure out which emergency alerts are relevant to specific geographic
areas.
Once again, the cellular industry is an example of fighting something.
Although most smart phones have built in mapping services and cellular
advertisers can track your geographic location within 10 meters; cellular
providers have been dragging their feet in geo-locating emergency alerts.
Automatically geo-locating indoor smart speakers and smart TVs is more
difficult, but if advertisers can get geolocation information from
AT&T, Amazon, Apple, Google, Sprint, T-Mobile, Verizon, etc; why can't
emergency alerts?
> Whether it's a good idea or even necessary to make those "helpful" features
> mandatory is perhaps a good question. At this stage, I'd probably lean
> toward no and see whether things resolve themselves on their own. The
> Internet, et. al., is pretty good at adapting to use cases like this without
> heavy-handed intervention it seems.
Sitting on the sidelines means you won't get to provide input on how it
works.
If you want to it to happen a specific way, then you need to contribute
that way early in the process.
The myth of the Internet figuring everything out isn't really true. Many
of the things people take for granted, are the result of some intervention
and even government requirements. Government requirements are often just
public intervention when industry fails to act on its own.
More information about the NANOG
mailing list