Rachel9 wrote:
He did test on multiple connections.
He tested two home users in the same geographical area with the same ISP. He contrasted that to the service from a business in the same geographical area. What he didn't do was test home users with a different ISP both in the same and different geographical area. Because that would tell him if it was something localized to the home VZ network, or something unrelated.
His tests don't actually tell us anything other than that during the same time period, two different home users, in the same geographical area, and using the same ISP, had similar network performance.
Most significantly, is that he didn't perform any tests at all to other remote sites. If one were to test whether this was VZ throttling network speed from/to a single specific remote service provider, you'd think that would be the very first test you'd do. Gee. I'm getting super slow network to AWS. Let me test the bandwidth to somewhere else. If it's slow *everywhere* then we can conclude that his internet speed sucks, but that it has nothing to do with targeted throttling.
But he didn't do that test, did he?
Quote:
I'm fairly certain amazon doesn't prioritize Verizon's business users over home users...
I could spend a significant amount of post time explaining how various internet companies do actually respect the others business agreements and pass them along (for a fee, of course), but it's a far more complex subject than you probably want to read about. However, in this case it's probably not necessary as the most likely explanation is that VZ prioritizes the traffic of their business accounts. Again, the red herring here is the assumption that this has anything at all to do with VZ to AWS. That's the connection he tested. But, as I pointed out earlier, he failed to test to anywhere else, so we have no way to know if his claim about throttling by VZ of traffic from/to AWS is true.
Quote:
...so his business connection is just as good as any other to make sure aws isn't completely broken.
Again though, that's not the question. He showed that the VZ business account got better performance than a VZ home account. To which most sane people would respond with "duh!'. The difference in the time it takes google to load up on my computer at home versus here at work is massive and very noticeable, yet I've never once concluded that this means that my home ISP must be throttling google.
Smasharoo wrote:
Because he shares a loop with like 1000 other home users.
Nope. Awesome guess from your time machine, but it's still not 1987 here in our reality. The service in question is dedicated last mile fiber.
Uh... Which is meaningless (and honestly a waste of money). I can wire direct fiber to 1000 homes, and plug them all into one set of switches, and guess what? They're not going to get any more performance out of their fiber than they would have with cat5. What matters is not the bandwidth of the wire from your house to the switch, but how many homes are wired into the same cloud of switches and routers before they fork into a backbone. Well, that and a host of considerations based on the layout of said cloud, of course, but that's a whole different ball of wax.
Having "direct fiber to your home" is the latest marketing scheme, but doesn't at all guarantee any increase in real connection speed to the internet as a whole. Oh. And you do get that the term "last mile" doesn't actually refer to a physical mile of distance, right? Also, it's not "dedicated last mile" at all. If it were, we'd be talking about wiring fiber directly from just your home right into the providers backbone. No one does that. Not for home users, at least, because it's massively expensive. Home users are clustered into a local switch cloud, which ultimately pipes into their backbone, but it's always going to be a balance between the cost of providing the pipe and the number of subscribers in the area.
Um... They do do this for business users, if they pay enough money, or sometime just because of where they are located. There's usually a much higher bandwidth capability and fewer bottlenecks between the physical wires laid out to commercial parks than to housing areas. If your business is in a commercial area with a ton of other businesses which have paid over time to have very fat pipe running right next to their offices, even the cheapest business account in that physical location will get excellent performance relative to a home user.
Edited, Feb 12th 2014 6:32pm by gbaji