So you think you have high speed DSL

[FreakinDJ]

So you think you have high speed DSL


They promised you a high speed connection, 1.5Mbds perhaps. It certainly seems faster then dial up service (56kbds), but then again is it as fast as they agreed to provide you.

Or perhaps you are one of the many Internet Savvy Surfers and you have tested your connection. Some companies at the time you establish your account will direct you to their web site to record your first speed test. Blazing speeds are immediately registered and another happy customer (Sucker) is on their way.

One of the fastest growing internet providers SBC Telecommunications is actively trying to rewrite the definition of Download. The trend is picking up momentum and many companies are establishing newer standards for what you are paying for. “Through Put” and “Transfer Rate” are being added to explain better what they are willing to provide you. This is most likely completely different from what you thought you were getting when you agreed to the terms and conditions of your contract.

First let’s examine “Through Put”. This is what the company techs will explain is the speed with which you load up web pages. Now remember you tested this speed and it is fast as advertised, or is it?

You need to first check the speed with some one other then your service provider. Service providers have loaded the internet with many of their own Speed Test Web Sites as will as many Pseudo web sites offering free speed test. Very simply when you go to their web sites it will Appear as if you indeed have the speeds you are paying for.

First try downloading this 15 day evaluation of Network Probe http://objectplanet.com/Probe/ This software will accurately asses the internet speeds you are actually receiving on your computer. Not only that, but more importantly display the protocols and IP packet size. Now back to the speed test.

Speed Test the consist of nothing more then 64 bit packets in no way can offer a true representation of network speeds. You simply do not load pages with 64 bit packets nor transfer files with 64 bit packets. It is not much more then a ping. Pinging your connection is not going to demonstrate your internet speeds, but it sure makes your ISP look good.
Then try goggle searching “Internet Speed Test” and sift through the many ISP company sponsored sites to find a true accurate assessment of internet speeds. Remember the one SBC has directed me to simply send out 64 bit packets to determine my speed. I can ping at speeds in excess of 1Mbds but can only load pages at 150Kbds.

Try your speed test here http://www.numion.com/

Now why is this an accurate speed test and why it is so much lower then the rated speeds I am paying for? Simply because these speeds are from reputable web site servers of extremely good quality all over the United States. Also and more importantly they are not Pinging your connection but sending a blend of IP packet protocols that accurately determines the rate at which you can load pages.

If you are unlucky enough to complain to your ISP and simply ask to receive what you are paying for, be prepared! You are most likely going to start experiencing billing problems, service disconnections, and unexpected service visits from your provider.

Document every thing. The network probe is a true representation of what you are actually receiving and can not be argued successfully by the ISPs

More to come
Stay tuned

[Don Claybrook]

1. Several issues at work here

Most all ISP's only guarantee a speed from you to the first hop, a router in their network, the first device your packets arrive at after they leave your home. Then you have to consider the ISP's oversubscription rate. How many customers do they subscribe before they get another T1 (or equivalent) from their service provider? This is the area in which most ISP's shirk their customers. After the packets leave the ISP, that ISP has little or no control over what sort of throughput you get. Of course, it would behoove that ISP to do business with a top-tier provider, but beyond that, and infrastructure issues such as redundant connections to different upstream providers, they can't really control anything. Also, the download speed you see in the lower-left corner of Internet Explorer doesn't take overhead into account. There are several layers of various encapsulation below what you see. In other words, transferring at 150K per second in Internet Explorer takes a good bit more than 150K of bandwidth. There are also retransmissions for data that was received garbled. You also have to consider remote-end issues. Does the remote network have a Packet Shaper, or some other device restricting bandwidth? Is that server overloaded? Is that server being served by a lousy ISP?

In the end, you can really only look at the first few hops of your connection, and primarily just the first. Once a packet leaves your own ISP, there are no guarantees.

[FreakinDJ]

3. Thats why an accurate assesment is needed

Servers such as ATT, MSN, Microsoft, and Yahoo are some of the best connected and most reliable systems out there. The same fiber network that carries your telephone conversations, internet and many digital Network syndicated television broadcast all end up in the same Central office. I know I have worked in them. So blaming the server/host doesn’t come into play here.

Prioritizing IP packet protocols and Collision rates do come into play. Mainly the ISPs are slowing down your connection to certain protocols. Collision rates are determined by the physical make up of the connection to your location. So where is the Through Put anyway? Through Put is not made up of 64bit packets.

[Don Claybrook]

6. True, 64K packets are the domain of ICMP, for the most part

I've worked in the CO space too. In short, I don't think you'll ever see MPLS for the end-user. Think about it: if you want end-to-end guarantees, you have to own all the parts in the middle. We don't want this, truly we don't, for it would mean one company (or more likely, the government) owned the infrastructure of the Internet, at least in the US. My nightmare is Ashcroft's dream. No thanks.

If you're talking about truth in advertising, I guess that's a different matter. Maybe someone could come up with new standards for measurement, but you're still not going to be able to guarantee much past the first hop. I still don't understand what you mean about server/host not coming into play. I mean, you can, as I'm sure you know, do traceroutes and find out where the bottleneck is. It's typically at one end or the other. As to restricting protocols, I'm against it and I believe ISP's should have to advertise such policies up front.

[BigMcLargehuge]

38. either owning the intermediary parts...

or having an SLA with a provider. Neither of which are within the domain of consumer networking technology.

[bowens43]

8. Not entirely correct.

The server , host, transmission routes, number of hops, traffic loads etc all play a roll in this. You don't have a fiber connection from your PC to the server you're trying to access. When a consumer purchases a DSL connection the advertised speed is the line speed from his location to the first server or to the DSLAM. That's it. Nothing more.

ISPs don't advertise through put or transfer rates.

P.S. We seem to have a lot of telcom people here. I have 22 years in the industry working for major carriers as a technician and engineer.

[Don Claybrook]

10. I'm a data guy who shifted toward telcom

...with a data emphasis. Lately, I've shifted more to unemployment. :)

I used to work for Williams Communications (was called WilTel at that time) as a "data guy". Our slogan was something about the convergence of voice and data. It was on all the stationery, on the website, on the on-hold message. And it was all bs. The two worlds never quite coalesced, at least in those days.

Later on, I worked for a company where my job duties included, among other things, setting up colocation equipment in 22 different AT&T POPs. That was fun stuff. More recently, I worked for a smaller VAR and ended up doing lots of T1 and frame-relay customer premesis stuff, and a bunch of ISDN and even some switched 56 (yuck).

So I can't really claim to be a telcom guy, but I've spent some time in the space. Nice to make your acquaintance.

[bowens43]

15. Hello....

I've been with MCI for 20 years and luckily have been able to dodge the bullet so far. I've always worked with transmission systems. Rumors are another big layoff in the next 2 weeks....oh well business as usual.

[ProdigalJunkMail]

29. Don, are you in my class this week???

funny got a guy in class with your name and the same scenario!!!

TheProdigal

[FreakinDJ]

11. So then maybe you can explain to me

and i don't want to get angry or into a pissing contest.

My experience is mostly in LAN so I do not qualify against your knowledge.

But why do the Telecoms use the smaller 64 bit packets for their speed test.

The technical department of SBC does say they advertise 2 things Through Put (the rate at witch you load web pages) and transfer rate (the rate at which you can download music)

I have proved to them over and over again that I am not receiving the speeds advertised. Ever since I contacted the Public Utilities Commission concerning this they have been working overtime Trying to correct the problem.

My concern is the deceptive means by which the direct customers to these speed test sites that use 64 bit packets.

THAT IS OBVIOUSLY DECEPTIVE

[bowens43]

14. The packet size is irrelevant.

Usually the advertised speed is the speed in bps from the customers location to the first pop/server/dslam etc. Usually when a customer complains of slow speed on their DSL connection the help desk person or technician will have them FTP a file from the first server they hit and then will translate the transfer rate to bps. The connection, whether fiber, dsl or analog is a serial connection and all speed measurements are in bits per second, there is overhead for the protocol that is also transmitted down that line. Usually if your transfer for a file is 85% of the advertised rate (or there abouts) your DSL line is functioning correctly. Maybe I don't see it as deceptive because I am familiar with the technology. I guess the average consumer may not understand.

[FreakinDJ]

24. I can garuantee you

They are prioritizing packet protocols on their network routers. We have an FTP site and the downloads are restricted to less then 100Kbds. We have been screaming at them for quite some time about this.

Then the harassment started. Our phone shut off, then on, then restricted. Our account numbers changed, notes removed from the file, balances shifted. So far the emails are angry now that the PUC has gotten involved

[ProdigalJunkMail]

28. yes, they probably do and are well within their rights to do so

otherwise, everyone and his brother would be putting up FTP sites and you think the net is clogged now??? What a mess that would create. It is bad enough with all of the file sharing that is happening now...there is a finite amount of bandwidth out there!!!

TheProdigal

[bowens43]

2. Not sure that I understand your complaint.

The guaranteed speed is line speed. That's from your location to the the first server. Any test to one of the many 'speed test' sites is not a valid test of your access speed. The service provider has no control over the route that is taken to reach this server. The most valid way to test your access speed is to ftp a file from a server provided by your ISP. The transfer rate should be about 85% of the guaranteed speed.

[FreakinDJ]

4. It is really this

A 64 bit packet is not much more then Pinging the connection. It by no means represents Transfer rates or Through Put. Most consumers are not aware of this.

when they go to the sites directed by the ISP they are being fooled into believing they have the advertised speeds. This practice is designed by nature to fool the unsuspecting consumer to say the least

[bowens43]

5. I disagree.

They probably do have the advertised speeds. ISPs are not advertising through put or transfer rates. I don't believe that anyone is being fooled. That implies deceit. The consumer needs make an attempt to understand the nature of the internet and the nature of the technology. Any test to any location beyond the first server is invalid. We can't really blame the ISP for the lack of knowledge on the consumers part.

[FreakinDJ]

7. That does not take into account

The physical topology of the SBC network is Fiber couple together with ATM protocols that FAR EXCEEDS the rated 1.5 Mbps max and 786 Kbds min speeds of my connection.

Actually they have no problem recognizing they are not providing me with the advertised speeds. They where out here working on the connection on a Sunday. Then again that is physical properties of the connection and by No MEANS addresses Ip Protocols being prioritized at the network level. Then again thats is SBC's network

Your arguement has one very important flaw. It does not take into account that 64 bit IP packets pass just fine from any location in the United States at the rated speeds. Packets of 512 bits (60 to 80% of web pages load in packets of this size)and above travel at a much slower rate

So if you don't receive packets larger then 64 bits at the rated speed then what are you getting

[bowens43]

9. Now that's different

you have a problem on YOUR line. That's entirely possible but that doesn't translate to ISPs are ripping people off by advertising one thing and giving them another. The advertised speed is on the access line only at bps rate.

[FreakinDJ]

12. Oh yes it does

They were satisfied to direct me to a speed test site that only used 64 bit packets. Had i not checked i would have never known the different IP packets were being prioritized at network router levels

[Fescue4u]

19. Not really

" Packets of 512 bits (60 to 80% of web pages load in packets of this size)and above travel at a much slower rate"

512 packets have much less overhead (% wise) than 64 byte packets. However 512 byte packets a just a bit longer serialization delay.

large packets are most efficient for transfering graphics files etc because the overall transmission will have less overhead.

small packets are best for real time apps, like voip because of the smaller serialization delay.

All in all. 64byte packets underrepresent your throughput, because a large % of the data transfered is dedicated to header overhead.

[spindoctor]

20. packet size

This is not entirely my cup of tea (plus I come here to discuss politics), but isn't packet size determined on the client side?
A website can't "tell" you what size to use.

[FreakinDJ]

22. Ip protocols

I believe they use "choke packets" to determine at what rate you can receive packets. The different protocols determine involved in the OSI layers (applications software) determine the packet size.

[ProdigalJunkMail]

27. IP packet size is determined by

the physical media to which the transmitter is attached. Those packets may be fragmented in transit due to smaller MTUs (maximum transmission units) and then reassmbled by the receiving station.

TheProdigal

[FreakinDJ]

17. Since you have so much experience

Maybe you can explain to me what does use 64 bit packet protocols

[ProdigalJunkMail]

25. nothing uses 64 bit packets

now 64 byte is another issue. There are many protocols that use them depending on the fuction they are carrying out. Ping is a good example...but not commonly done by home-end-users as they don't really know what their function would be. Essentially, a lot of ICMP functional protocols use 64 byte packets.

HTTP (for web browsing over IP) frequently uses small packets, a little larger than 64 bytes, to perform requests. When you download a page the workstation makes the request in usually very small packets and when the server responsds with the page requested larger packet sizes are used.

The description above of the packet overhead issue on small packets is correct. Essentially the rule is: the smaller the packet size, the more overhead involved. The reason being is that IP packets on any network have a fixed header size. For instance, ethernet has a 18 byte phsysical header requirement, the IP protocol has a fixed size (for all practical purposes) of 20 bytes and then TCP also has a fixed header size of 20 bytes...and you haven't even hit the data yet.

When it comes to download speeds, you really cannot count on guarantee with web service. There are just too many factors in play. I can tell you what the transfer rate to the DSLAM is or to an internal device but once you leave this network and route somewhere else...all bets are off...

TheProdigal

[Beaker]

13. My SBC DSL test numbers at numion.com-

Throughput Average: 51021 Bps 408 kbps

I haven't had dial-up for a few years, and I recently switched from cable to DSL. As to the speed- It seems fine to me, and I like the service a helluva lot better than Comcast's crappy cable service.

[FreakinDJ]

16. Can i ask what plan you have

My SBC DSL plan gives me a max speed of 1.5Mbds and a minimum speed of 786Kbds

My speed test this morning shows 338Kbds a far cry from the speeds SBC has guaranteed me in this plan. If I was receiving 80% of what was advertised I would have no grievance with them at all.

Yet according to the speed test conducted at their site, I was receiving in excess of 1 meg

It is the deception of using a Ping and referring to it as "Through put" that bothers me the most.

also when the novice orders the plan, they referred to it as "Download Speeds" Once you begin to question it, then the new language of "Through Put" and "Transfer rate" start to come into play. The true test is when you examine it and then when you see they only use a 64 bit IP packet.

[Fescue4u]

18. 64 byte packets are very good way to test speed

Those tax the network the hardest and will tend to underrepresent your actual speed more than anything else.

[FreakinDJ]

21. not in my case

The 64 bit packets travel at lightening speeds like 1.025 Mbds. the only problem is nothing really uses 64 bit packets. Things like web pages load in 60% 512 bit packets. Actually you will see 80% of the information you see on your screen is from 512 bit packets.

Try down loading the network probe. It is a real eye opener. Run the many speed test out there on the net and then see what you are really getting

[FreakinDJ]

23. You think I am wrong then I challenge any one

You think I am wrong then I challenge any one

To simply load the Network Probe found here http://objectplanet.com/Probe/

And then simply start conducting your own speed test. Try running goggle search of “Internet speed test” then look at the results you have. Notice very carefully that many of them have links and or are affiliated with the major ISPs. Then try running the test at http://www.numion.com/

Now what could be a more fair representation of true internet speeds? These are the logo icons from many of the most popular sites on the net. Then just to top it off try calling your service provider. Their technical support staff will most certainly direct you to their own site that will have blazing speeds

[ProdigalJunkMail]

26. DJ, you are missing something important

When someone tells you what their MAX and MIN transfer rates are, they CANNOT give you numbers for anything off of their network...cannot be done and any attempt to do so would be BS. An internal network server in SBCs net will give you the numbers they advertise because it is IN NETWORK. You leave their network by so much as ONE hop and they have no control and therefore cannot guarantee speed from that point forward. It is all a matter of where the network bottlenecks are and SBC has little to no control over external bottlenecks.

Also, you're just not right about packet sizes. The larger the packet, the better the throughput due to a lowering of overhead. Due to the way network switches and the phsyical transmission media works, it is MUCH more efficient to transfer data in large packets rather than small.

By the way, I work with protocol analyzers every day of my life. You want to test my knowledge...I am here to serve.

TheProdigal

OnEdit:can't type worth a flip

[FreakinDJ]

36. Well thanks for the offer

I know they can not guarantee the condition of the Internet once it leaves their network, but I am only looking for reasonable transfer rates.

It started out at 194kbds for this connection. Then we started of with calling their technical support people. We pretty much were fed up with what they were trying to tell us pretty quickly. They did double talk the issue and lost the argument pretty quickly.

After enough complaining we finally got fed up and sent the results of our network probe analyze, and speed test to the Public Utilities commission out here in California. Things started happening pretty quickly then.

First the Tech came out to our house and ran the SBC speed test from my computer. The results were impressive 1.025Mbds, put then he tried down loading some shareware and it was less then 100kbds. Then I pointed out he was only using 64 bit packets in the speed test from SBC

Will things got pretty hairy after that. All of a sudden I have these huge billing mistakes from SBC. They change my account numbers on me, notes are removed from my account files, my phone is restricted then back on again. Some times more then once in a day. This thing is really getting bizzaro and I know I am sounding like some paranoid wacko, but it appears once I begin questioning the 64 bit packet speed test I have struck a nerve

[ProdigalJunkMail]

30. just when there is a topic I know tons about...

this happens...too little too late I guess!

TheProdigal

[ProdigalJunkMail]

33. well, damn

guess I'll quit watching this one...

one...last...feeble...kick...freakinDJ, where are you???

theprodigal

[ProdigalJunkMail]

35. ok i lied...where is freakinDJ???

last kick...I promise

TheProdigal

[lpbk2713]

34. I know that's right.

Story of my life. A day late and a dollar short. Good thread though. :thumbsup:

[KDLarsen]

31. This beats it..

.. a Danish ISP have decided to add a point in their therms of use for flate rate connections, that states that if you use your internet connection above the average, you will be subject to having your internet subscription terminated!

[ProdigalJunkMail]

32. well, that should work for them

as their restriction would keep pushing the average down until it reaches ZERO and then everyone would be above average, hence get their service cancelled and then said company can go out of business...what a bunch of twits. At least they could do a set byte rate and charge for overages...even that sucks, but at least it isn't stoopid :-) ...

TheProdigal

[FreakinDJ]

37. ProdigalJunkMail I need you

Please help me determine what is going on here. This thing is way over my head and SBC is torturing me now

[FreakinDJ]

39. Damn just when some one was going to school me on this

Day late and dollar short the story of my life

[ProdigalJunkMail]

40. you rang???

TheProdigal

[FreakinDJ]

41. Sure did

First it started out at 192kds That was with the link for the speed test I provided. I thought that was accurate because it loads icons from 25 sites simultaneously. Kinda like jamming the bottle neck. Only this time the bottle neck would be the SBC network.

Also my kid has an FTP site. He tells me he never gets transfer rates higher then 150 kbds.

But as far as I can see web sites do not load pages in strictly 64 bit packets. SBC directs their customers to the SBC speed test that uses only 64 bit packets. Now isn't that deceptive at least

[ProdigalJunkMail]

42. brief attempt at WHY on the 64 byte packets

While the 64 byte packet is terribly inefficient at moving actual user data, in the terms of raw bytes (including the headers) it can be advantageous to use them for speed tests. The reason being is that ROUTERS (not layer two switches) are required to buffer the entire packet and check its integrity before forwarding. This means that when using smaller packet sizes the latency, or time spent languishing in buffers, is minimal. So, on an internal network, one that is routed much like SBC's is, small packets are good for overall bytes per second transfer. This is not what the real world does when moving user data, however.

The problem that you run into here is a bit of a marketing issue and unfortunately SBC is not lying about the connection speed that they are offering you. Technically, when offer information for connection speed, they could be speaking purely on the size of the pipe, meaning the MAX bandwidth EVER available on the link, and still be legally correct.

Believe it or not, 100kbps is not a hideously slow link. You must keep in mind that there are literally dozens of points of latency in the network and very few of those are within SBC realm of influence. The fact of the matter is, you could have the local connection speed of 45Mbps and have your own direct peering onto the internet backbones and still end up with transfer rates like the one mentioned above.

I know it can all be pretty confusing but unfortunately they are not really doing anything wrong besides marketing with slightly misleading terminology and not having the appropriate technical expertise to describe what is happening.

I hope that doens't make it any more confusing...

TheProdigal

[FreakinDJ]

43. Thanks for clearing that up

Really appreciate the explanation to me.

Can you explain on more thing with the expertise you so eloquently wield.

Why do they differentiate in between transfer rate and through put. Now their definition for through put is the rate at which you can down load a web page. They tell me their network is designed to download the web pages faster. The transfer rate is what they describe as the rate at which you can down load a file such as music. Is the router prioritizing?

[ProdigalJunkMail]

44. technically misleading

use of terminology. While the terms are used in their proper context, most of us really don't distinguish the difference. Essentially I don't care about the difference as long as I can download my music and my web page load times are quick. When it comes down to it, a byte is a byte and the time it takes to transfer it across the internet from origin to destination is what I really care about. The problem with that is that there are tons of place where that data can be buffered, redirected, held up and even lost. Are they doing traffic prioritzation? Probably. Traffic shaping can be done on several different criteria and one of those is certainly protocol type.

The end all beat all of it is that SBC is not lying about the connection speed they are giving you. If they can, at any point, get data INTO your machine, from ANY server, using ANY protocol then they are meeting the legal requirement of the agreement for your high speed service because they are giving you a pipe size (based on XXXKbps). How that pipe is utilized, be it efficient or not, is NOT what is being sold. I bitch and moan about this too, but we have to realize that many times it is beyond the control of our ISP.

Sorry to be sooooo longwinded, but I frequently teach this stuff and it just sort of 'gushes forth' once I get going!

Good luck!
TheProdigal