Verizon Public Policy on Netflix

Mon Jul 14 20:16:56 UTC 2014

The choice for ISPs at larger scale is peering or caching, peering is cheaper than caching as power is not as cheap as you think as well as the requirement to have two of everything for failover if you do caches (ie can't have my transits or more likely my backhaul blow up if the caches go away).   

I also typically don't want to give up the opportunity cost on the power in our main pops as it is not what the power costs, but rather what you could sell it for that matters in most of our core sites.   We don't cache in head-ends as we still would need the backhaul anyway if the caches fail so we can't really reduce the backhaul requirement much.  We have some middle tier sites in the cable network, but the benefit of throwing caches at those locations has never really been there since they are not staffed the same way etc.    I think a lot of big networks have this issue.

John 

-----Original Message-----
From: NANOG [mailto:nanog-bounces at nanog.org] On Behalf Of mcfbbqroast .
Sent: Monday, July 14, 2014 1:09 PM
To: nanog at nanog.org
Subject: Re: Verizon Public Policy on Netflix

I do agree that Netflix could offer caching services for smaller ISPs. But that's a fight for another day, right now were focusing on whether Netflix should pay for caching content, let's look at the cost comparison.

NOT CACHING with Netflix
- up to 8gbps of transit - what's that, several grand a month from a major hub with a big commit?
- a 10gbps port to transit provider

CACHING with Netflix
- up to 500w of power and 4u rack space - in a commercial DC that's a few hundred a month, most telecoms have rack space in their own office
- a 10gbps port to server - the same
- transit commitment in off peak hours - most telecoms have plenty of this to spare

That's a pretty massive saving.

I still do not understand how Netflix should pay for customers using your network. Its like charging another carrier to receive a phone call from your network, because you want to have cheaper plans.

The risk is, the policy Brett suggests, will misrepresent ISP pricing. This is a huge issue. Brett? How do you think you can compete with big providers when they're subsidized by Netflix? Bare in mind they'll have much more power in negotiating with Netflix than you. Your customers will be paying for Netflix, subsidizing your competitor!

Finally, I'd like to point out that there's an ISP in New Zealand called slingshot that popped up on my radar. Transit in NZ appears to be expensive as hell ($20+/Mbps for bulk buys from competitive PoPs) yet this ISP, Slingshot, encourages customers to use their VPN to access Netflix.

This is notable to our conversation because when any ISPs are proposing whats essentially a "Netflix tax" another one, who pays 20x or more for transit and cannot cache Netflix are encouraging use of Netflix. Why?
Publicity.

Brett, you might like a look at that because they charge $10 more than the cheapest competitor, but the proxy service they provide (which probably costs them pennies) keeps customers flowing like water for its ease of use.
In a age where internet is becoming a commodity these are the types of services that can keep you afloat.

Alternatively, use this debacle as advertising! I've seen many cable users complain about Netflix being very slow, could advertising that you don't throttle Netflix give you a competitive edge in cable territory??