General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsVerizon ALREADY throttling Netflix, Amazon
Verizon Using Recent Net Neutrality Victory to Wage War Against Netflix
Dave's Blog-
I usually dont post articles about current affairs. However, a recent series of events has inspired me to write about this.
Towards the end of January, the president of our company iScan Online, Inc., was complaining that our service was experiencing major slowdowns. I investigated the issue, but I couldnt find anything wrong with our production environment. We were stumped.
One evening I also noticed a slowdown while using our service from my house. I realized that the one thing in common between me and our president was that we both had FiOS internet service from Verizon.
Since we host all of our infrastructure on Amazons AWS I decided to do a little test I grabbed a URL from AWS S3 and loaded it.
40kB/s.
WTF.
I also noticed that our Netflix streaming quality is awful compared to just a few weeks ago.
more
http://davesblog.com/blog/2014/02/05/verizon-using-recent-net-neutrality-victory-to-wage-war-against-netflix/
octoberlib
(14,971 posts)Is slow for me and I have FIOS - I thought it was just my cheap laptop acting up.
Fios is pushing their redbox service - I think they want people to quit netflix and order that.
Egnever
(21,506 posts)his tracerout shows it.
Aerows
(39,961 posts)shows connection speed, and you are claiming someone else is an idiot?
The traceroute clearly shows the problem starts with his router.
The traceroute shows response times from each hop along the path. Starting with HIS router. Notice his home router has clean response times of 1ms
but his business router shows ridiculous 18.036 ms 1.326 ms 2.318 ms response times. His problem is his router.
And yes I am claiming once again this fool doesnt know shit about networking and neither apparently do you.
Aerows
(39,961 posts)on this topic before leaping in and calling those in the field fools.
LOL. "Tracert told me everything". Um, yeah.
Egnever
(21,506 posts)DisgustipatedinCA
(12,530 posts)Please state your credentials, because I don't believe you know what you're talking about.
Egnever
(21,506 posts)and I may be misreading the traceroutes but the first does not give response times in decimals. and the second apears to be giving response times win the thousands range. If he did different traces I didnt catch it and I am indeed the idiot.
But then if the times are in decimals then the traceroute proves absolutely nothing. Because neither of them are bad in fact his business one would then be almost identicle to the home one except the small blurb in hop three that would be nothing to complain about assuming it is 67 ms and not 67,000 ms.
Paulie
(8,462 posts)Given no change in the tracert they are probably doing some DPI for certain prefixes or AS' and throwing it Into scavenge or reducing the MSS and/or window. Lots of dirty tricks to be done to get you down to 50Kb on fiber.
Aerows
(39,961 posts)But I can't.
What is this I dont even
JDPriestly
(57,936 posts)our router. But the increase in bandwidth is due to changing our internet provider. It may not be just the router.
whopis01
(3,521 posts)The connection to Netflix was better from the business than it was from his home.
And yes, the home router had a little better response time - but wouldn't you think that would make the performance of the Netflix connection better at home? It was the opposite.
Why do you think a response time of 1.326ms or 2.318ms is ridiculous? Agreed it is a bit worse that the 1ms he was getting at home, but it isn't that big of a deal. Likely either a bit more complicated switching networking between his computer and the router at the office, or packet filtering happening on the router.
bl968
(360 posts)You seem to be the only one who doesn't know shit. As a Network Administrator for 16 years I back up the testing methodology used in this case. As for the 18ms on the work connection, it shows a loaded circuit no more no less. That being said is there enough here to make the claim that it's throttling, no; is there a reason to suspect throttling; yes.
n2doc
(47,953 posts)PoliticAverse
(26,366 posts)Egnever
(21,506 posts)The guys blog proves he is an idiot and has no clue what his tests show.
Aerows
(39,961 posts)and discover a Networking Essentials course. Or, you know, basic information on what console commands display.
Egnever
(21,506 posts)Aerows
(39,961 posts)You are using tracert to prove you are right.
DisgustipatedinCA
(12,530 posts)Starting at hop 4, the trace is identical. Starting at hop 3, both packets are on the same router, just connected to different interfaces. So what is it about the traceroute that bothers you?
Aerows
(39,961 posts)most routers drop big packets sent via tracert. for starters.
I could go on, but starting right there ... well.
Warren DeMontague
(80,708 posts)Of course, they swore up and down that wasn't why they fought it in court.
Egnever
(21,506 posts)he just doesnt know what his traceroute actually shows... Really really lame!
Warren DeMontague
(80,708 posts)But that said they are denying it and disavowing their own rep's statement on the matter, so there's that.
Aerows
(39,961 posts)used *tracert* to pretend to know a connection speed. Um, yeah. And I'm a resident of OZ and can tell you how much power the Wizard has by traveling the yellow brick road.
Warren DeMontague
(80,708 posts)Might be interesting pieces of data.
I'm not enough of an expert on networking to decipher them, personally.
Egnever
(21,506 posts)Warren DeMontague
(80,708 posts)All he says is he thinks they "might be significant".
Aerows
(39,961 posts)Tracert tells you that it gets from point A to point B. It tells you very little under such circumstances because a packet sent is so tiny compared to MB/s of data that it is worthless.
DisgustipatedinCA
(12,530 posts)It could also be significant if the round trip average was an unacceptably high number, and stayed that way in subsequent hops. No idea what the other poster is talking about re the "work router" at 192.168.1.1. I see a lot of latency add at hop 3,'but the average time drops back down after that. This is called ICMP throttling--when a router gets busy, it can choose to ignore traceroutes, pings, and so on, so that it can do its main job of forwarding packets. If the latency had gotten abnormally high at that hop, and stayed that way in subsequent hops, then one could safely assume that a slowdown was being introduced at that hop (but not necessarily a product of throttling). The router at hop 3 is a Verizon router, by the way.
What does the traceroute tell us? That both streams of traffic mostly go through the same equipment (beginning at hop 3). There could be a problem at the "work router" but I see no indication of that in this traceroute.
Egnever
(21,506 posts)and you dont see a problem with that?
DisgustipatedinCA
(12,530 posts)Have you ever tried pinging from a Cisco router to something on a local segment that isn't yet in the arp table? Do you know what the first ping looks like? A timeout...that's what it looks like. The average round trip time for 3 packets to the first hop is 1.2 ms.
Aerows
(39,961 posts)Last edited Wed Feb 5, 2014, 10:06 PM - Edit history (1)
ARP table replication is a pain, but you have to endure it putting a new one on line. It's like DNS servers and Active Directory. Prepare to fucking wait until everything gets replicated out and they all "find each other".
Aerows
(39,961 posts)I can pick up on a port problem but I'm not seeing it here.
Aerows
(39,961 posts)Ignoring ICMP traffic and plain packet traffic.
wandy
(3,539 posts)Tracert tells how you get from point A to point B and is NO indication of sustained packet flow.
1) Although HULU and not Netflix I have no problem with video normally 480p.
2) Were Tracert an indication of sustained throughout (it is not) I would envy his numbers.
I suggest he seek the cause of the problem elsewhere.
Aerows
(39,961 posts)in routers and judging packet size is part of this. It doesn't necessarily mean anything is flawed in the hardware, it doesn't mean anything ISN'T flawed in the hardware. It does make the network questionable, however.
wandy
(3,539 posts)The bottle necks that can occur within the PC are many and diverse.
Although simply playing a DVD may tell the tail I have a quick test for video performance.
Believe it or not a YOUTUBE Video.
Collings B24J Ball Turret.
Using any of a number of youtube "downlowders" it can be obtained from 144p up to 1080p.
Once it is on disk you can get a fairly good idea of the system response with no network involved.
If you do not get smooth video at 480p, decent streaming video isn't going to work out too very well.
At 15 minutes it will provide a reasonable indication of sustained throughput.
Aerows
(39,961 posts)wandy
(3,539 posts)More in the league of Electron Microscope.
Nice. Envy!
Business system is AMD 8150 4.3 ghz tops. And does a fine job heating the house while it's at it.
The A/V system (da lil guy)
AMD 6400 95W @ 3.6 ghz.
Asus M5A88-V-Evo
8 gig main @ 1600
120 Gig SSD for Res/App/paging/Video-Audio buffer
M-Audio 2496
All fans, Magnetic Barometric.
And, don't laugh, ASUS 8400 Silent for video. Yup Geforce 8400
1080p. No problem.
By use of various A/V formats it has taken on the job of VCR,DVD and CD decks.
Windows Media Player + K-Lite = Ya can deal with anything.
Much Vinyl has been preserved, still, NOTHING sounds like the real thing.
Room cooling is a problem as it shares space with 3 dual mono power amps.
I've had some health issues so I haven't been 100%, but I can deal with that as a nuisance.
Aerows
(39,961 posts)Ya'll need Tesla. Ferme. and Cuda kicking ass.
wandy
(3,539 posts)OK, theres better. I would like it but..........
Video cards are a thermal problem in and of themselves.
All amplifiers powered on, all channels driven, just short of 1000 watts RMS into 8 Ohm.
Nominal both pair (their ain't no real front and back) 6 Ohm.
Bass, nominal 4 Ohm.
Not loud, just accurate!
At least you can appreciate why a 95 thermal watt CPU was called for.
It is a computer living in a sound world.
No, not THX or Dolby 7.1 but closer to what this might have become.
http://www.widescreenmuseum.com/sound/fantasound1.htm
By the by. Glad you are OK. Nasty but you did the right thing.
That seems like it would be loud as hell to cool.
Thank you for your well wishes. They are appreciated.
wandy
(3,539 posts)No I'm not going to do the expense of fixing the "Whole House". No a window air-conditioner is just more buzzes and rattles to deal with. One of the few good things about retirement is not spending so much time on sinus rattling raised floors.
After a point, Audio becomes more subjective than anything else. No matter how much money you throw at it.
Once upon a time Disney might have been honest.
The depression, WW2 and later corporate greed may have gotten in the way.
Still, that funky mouse may appear in the audio image exactly where you perceive him to be.
You're just going to dump some excessive heath into the surrounding environment to do so.
Aerows
(39,961 posts)They are about 28 bucks and can cool the crap out of a processor if you have room in your case. They rock!
Aerows
(39,961 posts)I didn't know this was an issue. Take care.
Aerows
(39,961 posts)that clocks at 4.5 (it's a 2500K unlocked) for nearly two years, with a tower cooler. The whole system has been overclocked to hell and back, but all I've done is hyper cool with fans. Fans. I've racked up numbers on system registers like futuremark that are amazing. But they are just with amazing, select hardware, and great motherboards.
Egnever
(21,506 posts)but his traceroutes prove him wrong
Here is his home trace
Tracing route to iscanonline.com [23.21.158.115]
over a maximum of 30 hops:
1 <1 ms <1 ms <1 ms 192.168.1.1
2 7 ms 7 ms 8 ms L100.DLLSTX-VFTTP-65.verizon-gni.net [173.74.57.1]
3 10 ms 6 ms 9 ms G0-5-2-0.DLLSTX-LCR-21.verizon-gni.net [130.81.190.204]
4 16 ms 9 ms 10 ms so-5-0-0-0.DFW9-BB-RTR1.verizon-gni.net [130.81.199.34]
5 10 ms 9 ms 9 ms 0.xe-3-3-0.BR2.DFW13.ALTER.NET [152.63.100.5]
6 9 ms 10 ms 9 ms 204.255.168.158
7 10 ms 9 ms 10 ms ae-1.r08.dllstx09.us.bb.gin.ntt.net [129.250.3.27]
each line is a hop along the path to the destination the numbers with ms behind them indicate the time it takes for the signal to go from his computer to each step in the path along the way and back. He left off the first step of both patch that would be his own compter probably to hide his home IP adress. But the first jump listed is 192.168.1.1 whis is a very common default adress for a router. Notice at home when he takes that first step the response times from the router before he ever gets to verizon is <1 ms that is a good response and what you would expect to see. All of the hops continuing down the path have decent response times notice #2 that is where he first hits Verizons network and the times there are also normal in a system that is running properly.
When you look at his business though we see something completely different.
1 192.168.1.1 (192.168.1.1) 18.036 ms 1.326 ms 2.318 ms
2 l100.dllstx-vfttp-93.verizon-gni.net (71.244.30.1) 5.870 ms 5.211 ms 5.193 ms
3 g0-5-0-2.dllstx-lcr-21.verizon-gni.net (130.81.138.12) 7.400 ms 67.679 ms 10.605 ms
4 so-5-0-0-0.dfw9-bb-rtr1.verizon-gni.net (130.81.199.34) 12.062 ms 6.652 ms 17.799 ms
5 0.xe-3-3-0.br2.dfw13.alter.net (152.63.100.5) 7.207 ms 7.858 ms 9.616 ms
6 204.255.168.158 (204.255.168.158) 7.435 ms 7.256 ms 10.366 ms
7 ae-1.r08.dllstx09.us.bb.gin.ntt.net (129.250.3.27) 7.365 ms 10.160 ms 9.083 ms
Notice the first step again with the default router adress 192.168.1.1
This time the response times are in the thousands of ms (miliseconds) anything after that is going to be screwed regardless because the router itself is broken. His problem at work starts before he ever hits the verizon network.
If he unplugged the router from verizon and just pinged the router itself he would still have horrible response times because the router is malfunctioning.
Aerows
(39,961 posts)when utilizing primitive tools to demonstrate network connectivity. Modern routers will routinely drop certain small packets that aren't endpoint packets. If you are "Hi, I'm Joe" at the console, you get a far lower prioritization than "Here's the spreadsheet you needed". Why? Because "Hi, I'm Joe" inconveniences Joe for about a 1/4th of a second. Inconveniencing Jill, costs about 3 minutes.
This is called prioritizing traffic. It can make things difficult to diagnose when not endpoint to endpoint, but there are a lot of better tools to come to the same conclusion.
Verizon is screwing their customers. I regularly deal with AT&T, and before them, I dealt with Qwest.
Egnever
(21,506 posts)spirald
(63 posts)The largest number in this traceroute is the odd 67.679ms (approx. 67ms). Aside from that outlier, the business trace looks marginally better than the home trace.
The business trace is providing resolution in the 1/1000 of a ms (i.e. microseconds) which is why there is a decimal point in the business trace. The home trace only provides millisecond resolution.
It looks like you are interpreting the decimal point as a comma, and that is why you think the work numbers are a thousand times higher than they are.
That was my mistake and I retract that his router is bad. However given that, then the trace says nothing as the numbers are pretty similar.
So that leaves us to relying on what a phone tech says to get a guy off the phone to claim they are throttling it.
I am not saying that they aren't they may be but what this guy gives as evidence is pretty thin.
I do notice he updated the article with a link to something that tests throughput to different servers.
I Think it's good that people are keeping an eye on this. It will clearly become widely noticed if in fact the companies do start doing this.
And it wont be a phone tech that tells everyone.
The Straight Story
(48,121 posts)I am guessing it is a very small business. I don't think any company I have ever worked for allowed use of that
Egnever
(21,506 posts)Paulie
(8,462 posts)Mostly for loop backs and point 2 point links. It's just a 32bit number after all.
DisgustipatedinCA
(12,530 posts)In this way, network people eventually always get the chance to learn double-nat/twice-nat to get around the problem of having two 192.168.1.0/24 networks on either end of a LAN-to-LAN IPSEC tunnel.
I'm selling blocks of RFC1918 addresses cheap. Get 'em while they last.
The Straight Story
(48,121 posts)They are considered 'consumer' ip addresses and companies using VPN's for people to connect to from home tend to shy away from using the same schema at work.
My data centers had over 5000 servers and I don't know how many devices off hand connecting to it (atm's, laptops, vpn, storage devices, printers, etc and so on) and none of the networks along the line internally used the 192.168 (except, as noted, a few closed labs for training and testing).
It is more of a 'best practices' in networking in businesses than anything.
Aerows
(39,961 posts)Aerows
(39,961 posts)because you obviously have no experience in Network Engineering. I notice you like to sneer at a lot of articles that criticize big business. Well, on this one, you came to the right place, my dear. Please educate us all on how a tracert illustrates network speed.
I'll wait.
Egnever
(21,506 posts)Aerows
(39,961 posts)but the first thing I would be doing is a netstat. Could you illuminate the rest of the class on why I would do that?
Cause you want to overthink it?
Pretty clear from his traceroute his router at his work is puking.
Aerows
(39,961 posts)that I never "Overthink" it when diagnosing network problems at work. I didn't realize that running mail servers for a 35,000 node network was a profession where "overthinking" things was superfluous.
closeupready
(29,503 posts)if they were unhappy with any single provider. So even if true, Verizon will be penalized by the free market.
photokat
(1 post)We live in a rural area and we don't have many options... there are three. Local ISP, Charter, and Hugesnet (sp?) the local ISP isn't cheap but we don't like Charter and Hugesnet isn't reliable so NO we don't have many options but it won't matter anyhow because all we will be able to get is Multi National Corporations websites unless we still have net neutrality....
Warpy
(111,317 posts)Everything goes lickety split during the wee hours of the morning, when sensible people have padded off to bed. It's one of the few benefits of having one's sleep patterns wrecked by decades of graveyard shift work.
Cha
(297,475 posts)my Netflix and amazon streaming is great. Knock on wood.
I'm behind.. has Verizon gone into the streaming business? I don't understand this.. but, am interested because I do a whole of streaming.
Vashta Nerada
(3,922 posts)Rozlee
(2,529 posts)But, I'm having a lot of problems with T-Mobile coverage here in the Texas Hill Country. I can't get any coverage when I'm at home unless I use their WiFi feature and stay indoors. And T-Mobile doesn't have a great record of clean business dealings themselves. I'm looking to switch to Credo if they'll buy out my contract. They claim to support progressive causes.
Vashta Nerada
(3,922 posts)T-Mobile's coverage there sucks, but I love the wi-fi calling feature. I stay in the house most of the time when I'm there so the horrible coverage isn't really that big of a problem.
Drew Richards
(1,558 posts)can denote ANYTHING for sure except route-path is a fucking idiot and should not claim they know anything about how the internet works...
Let me give you a little primer to get you started...
Traceroute as with ping uses a DEFAULT tiny little packet default for PING is 32 bytes....32 B Y T E S
Default for Traceroute is 64 B Y T E S...
Running a default "NO-LOAD" Ping or traceroute wont tell you shit except whether it is getting there and what route it takes...yes it might show some latency but does'nt really tell you anything about bandwidth...!!! a ping across the US on excellent networks usually is around 4-16 medium or busy networks 16-20 and latency is when you start seeing anything above 30ms for a ping across the US...your mileage may vary depending on if you have a backbone connection or not...
NOW...if you want to see what kind of LOAD/Bandwidth a path will accept...keyword here is accept...
You first have to do a traceroute and or a ping at say 1/4 256 1/2 512 1 MEG 1024 size to see if the route accepts it..in laymen terms...can the path hand NORMAL T-1 Packet Size of 1024 (1026 really now adays with all the ethernet) Or hey maybe it can handle a JUMBO packet of 9216? try running those and see how your routes look...
EVEN this really doesn't tell you (*^%% about the quality or reliability of the link...
NOW for some basic Network Education for the Laymen:
There are sites on the internet called...
LOOKING GLASS SITES and
ROUTE SERVERS
using THESE tools you can log onto the looking glass of verizon, bell south, AT&T ect... and run trace routes, bgp lookups...route-path consistencies SPEED TESTS!...ect...
On the ROUTE SERVERS you can do the Various sized packets to test reliability of the path and ability to pass traffic to your end point...
BUT WAIT there's MORE!
Now to really know WTF is going on you need to do a "BANDWIDTH/SPEED Test...
How you may ask?
Its really simple...
Either you
A. Know wtf you are doing and can send and receive a nice clean compressed digital file of say 10 meg to and from the networks you are testing and have installed the tiny little speed test script that is FREE to install...on a unix box...allowing others to also speed test off of you...
OR
You can do what most people that DON'T have the access to servers all over and use
www.SPEEDTEST.net....or any other speed test site...hell most providers have a link to their own on their LOOKING GLASS sites!!!
from your affected site go to speed test and PICK the sites/networks you want to run the test off of...and where you are going or to yourself...
Hell pick various sites and various carriers make sure its not YOUR network Provider gateway thats not screwing you...
a traceroute....a TRACEROUTE???...my head hurts....
(Forest Gump) And thats all i got to say about that...
Paulie
(8,462 posts)Icmp is good for verifying reachability buy nothing if there is an inline L2 DPI box tweaking the frames and adjusting the segment sizes or injecting bad bits into the payload.
That's what Comcast used to do before they got caught back in like 2009.
Drew Richards
(1,558 posts)immensely if nothing else to determine what it is NOT...
we used the same port based filter throttling when we set up various platinum gold silver
Plans for DSL customers back 15 years ago to limit certain uses on the internet...
We also set up caching servers to limit their outbound traffic usage requirements...I won't mention which carrier but it was not Verizon...but...the speedtest is a pretty reliable test of your raw through put...if you really want to go detailed then you are going to have to work PORT specific throughput tests but that is way beyond a layman on here to do or to test.
At that point you are going to have to work directly with your provider on port specific thoughput...IF they are willing to work with you. And the only way you can get that kind of assistance is to be a backbone/network engineer for a provider with justification for opening the ticket and having it worked.
But just a trace route...for diagnostics...PUlease...
Aerows
(39,961 posts)LMAO
JDPriestly
(57,936 posts)Verizon should be limited in its ability to control such a large vertical portion of the market. Companies should not be allowed to offer both cable and internet or both cable and phone or both cable and cell phone.
We buy our internet service from our cable company. We too have been noticing that Netflix is slower than it was a few weeks ago.
Personally, I think that if private companies are not going to compete fairly or are going to control our access to internet services so as to favor certain sites and disfavor others, we would be better off with just having a government controlled internet.
Fair competition is the advantage in capitalism for consumers. If companies gain a lot of control over, say the internet access services and use that to clobber their competition, then consumers are not getting the benefit from the competition that capitalism is supposed to provide. What is the use in capitalism if big companies squeeze their competition out of the marketplace?
This is something that the proponents of capitalism should be thinking about. Most Americans support the concept of capitalism. But is it capitalism when you have a product like bandwidth that is limited and companies use their control of it to force consumers to favor the company's products or services?
And if it isn't, then why not just have the government provide the product or service? The government might be more amenable to giving everyone the ability to buy what they want.
Who needs corporations and capitalism if the conglomerates use their money and power to control consumers' purchasing? I am really asking that question. I do not dislike capitalism, but if we don't really have the benefits of it anyway, what is the use?
Egalitarian Thug
(12,448 posts)As if having some of the least protection for workers isn't enough, no administration since Carter has enforced even those.
Egalitarian Thug
(12,448 posts)WillyT
(72,631 posts)blackspade
(10,056 posts)I'm not that hopeful though.
quakerboy
(13,920 posts)Hello, lets talk about what package I can offer you.
I buy the extreme basic internet from them. I pay the extra 10 bucks to do without cable tv that would cost me 30 bucks more.
Repeatedly through their conversation, they went back to how using their streaming version would have no slow loading issues, repeatedly asserting that it was better than netflix, specifically, in that way. I havn't seen any issues loading with netflix, but I'm sure that will begin presently.
Now that they have been enabled, there seems to me little doubt they will try to turn their new power to restrict internet access into a market advantage. My real question is once they have that all settled, how long it will take them to extend that power into restricting free speech, eg comcast locking out "comcastsucks.org", or venturing into throttling websites like our very own DU.
Mosby
(16,334 posts)32.6 mbps download and 8.6 mbps upload.
From my kindle it's about 19.5 mbps download. I have a wireless-n router.
speeds seem pretty good to me.