Netflix Isn't Swamping the Internet
itwbennett writes "Remember the Sandvine report from earlier this week that said Netflix gobbles up 30% of Internet traffic during peak hours? It needs clarification on a couple of important points, says blogger Kevin Fogarty. First, yes, Netflix traffic spikes during prime time, but only across the last mile. Second, ISPs underestimate what a 'normal' level of Internet use really is. 'When AT&T announced its data caps – 150GB per month for DSL users and 250GB for broadband – it called the data levels generous and said limits would only affect 2 percent of its customers. It turns out Netflix users take up an average of 40GB per month just from streaming media, according to a different Sandvine report (PDF).'"
And statistics. Even if it would only affect 2%, that won't be for long. They knew they had to put caps in now, because later it would cause too much backlash. Could it be that the "Internet" could be swamped by digital media? Perhaps, but they could always add more bandwidth. Although then that would hurt their earnings having to invest, much less being able to nickel and dime customers.
I wish companies like All-tel wouldn't have sold out. Though they weren't perfect, they had a lot going on right, and that is why they were successful. On one hand I am glad I am still with them, on the other, the rest of the family was moved to Verizon, eliminating one of the great reasons to join the same network.... But the big boys gobble up anyone that comes close to doing things right, so I don't see any reason to have much hope.
Like a city whose walls are broken down is a man who lacks self-control.
In my humble opinion.
He who knows best knows how little he knows. - Thomas Jefferson
So the last mile is the tightest, and contended. And we now know the data caps are a joke. So, still a problem.
This is because Netflix hosts their shit with caching companies. You get people like Akamai that do data hosting. Now they have big data centers that hold lots of data as you'd expect, but they also have cache engines all over the place. They contact ISPs and say "Hey, we'd like to put cache engines in your data center. We'll provide you all the equipment, free of charge, and tell you how to configure it. This will reduce the amount of bandwidth you use."
You can see why ISPs like this and go for it. Of course the other side of it, the reason Akamai does it, is because it reduces their bandwidth usage a lot. Win-win situation.
This happened on campus like 8 years ago. Akamai gave us some cache engines and they got set up on the network. Now anything on them is just stupidly fast. Windows updates just fly down. It also made quite a noticeable dent in off campus bandwidth usage.
I don't think Netflix uses Akamai themselves, but I do know they use a service like it.
"40 gig just from streaming data" with a lowball 150 gig allowance, with recent slashdot articles saying netflix is a large minority of people's traffic... sounds like the ISPs are correct, that 150 gig is generous.
slashdot: where everyone yells sarcastic metaphors to themselves to understand the issue
presumably, in poster's mind, broadband = cable. Maybe it means fiber but i would think cable. Most of the dsl implementations are pretty crappy around the states (based on setups around the east coast mainly, seattle was good). Because of this, most people think dsl is inherently inferior to cable broadband. having used some excellent dsl providers in london, it definitely comes down to the service provider.
Full disclosure: I used to work for a group within Comcast that looked at network traffic to the user. Let's just say I have a really strong dislike of all things Comcast.
With that said, not a chance that the ISPs are not estimating correctly. They aren't estimating. At least at Comcast, they have an incredibly good idea of how much network traffic is going through their system. And they build to a given percentile of busiest time in the entire month.
The only way you can say they are miscalculating what is going across the network is if Sandvine is not properly analyzing network traffic and is associating it with an incorrect network protocol.
Except for ending slavery, the Nazis, communism, & securing American independence, war has never solved anything.
So a company that sells network control and monitoring software, and who has a dodgy past, says the bandwidth caps are OK.
tomorrow who's gonna fuss
In the minds of most people, including most geeks, broadband = really fast Internet, and the cutoff for that changes year to year and person to person.
In reality broadband means, well, broadband, as in a service that is not baseband. So Ethernet, even 10G, is not broadband. However DSL, no matter how slow, is broadband.
Unfortunately, this shit is going to keep happening particularly now that the FCC has an official definition for broadband and it includes a minimum speed. People are going to keep misusing the term until the meaning just changes to "fast internet, where fast is whatever I think it is."
They don't buy, they rent from limelight and level3. They used akamai before that as a cdn (and I think might still use it some). Between these 3 the backbone infrastructure is pretty well covered. If they really do make up 30% of peek traffic it is reasonable to imagine that netflix has localized storage of at least the most popular content at most of these nodes, and the fact that limelight specializes in video transfer (even before netflix) makes it even more likely.
Get a web developer
I'm no lover of Comcast or AT&T, but I think the point about ISPs underestimating normal use is unfair.
It says that Netflix users take up an average of 40GB per month just from streaming media. In my experience, your "average" user isn't doing anything that uses more bandwidth than Netflix. Even with the lower data cap of 150GB, that leaves room for a three-fold increase in streaming bandwidth before you come close to using your allocation, with room left over downlaoding 3 or 4 full-sized games a month. Even with the supposed doubling of that rate for console users (which I doubt), that leaves plenty of room. And Mr. Fogarty needs to check his math, as 80 150*2/3.
Even if console Netflix users were averaging 100GB/mo for streaming, who can use 50GB/mo on email, web surfing, and youtube?
I think the author is overestimating how much bandwidth average users need.
Full disclosure: I am far from an average user. I have Netflix and DirecTV, both of which I use streaming video on. I also download a few DVD-sized images every month, and my wife practically lives on the web in the evenings. And yet according to my Tomato router stats, I've never even hit the halfway mark of my 250GB Comcast cap.