ISPs Offer Faster Speeds, Why Don't We Get Them?

Ron Williams asks: "I'm infuriated every time I see that companies are raising their speeds when they can't maintain their current speeds. Here's my biggest issue: my grandmother signed up for the 3Mbps DSL plan through Verizon, however a speed test said she was only getting 750Kbps. Why pay for the extra bandwidth when she's not getting it? She downgraded to the 768K plan expecting to still have 750K. Wrong, instead her speed dropped to 300K. So, how about instead of companies constantly claiming to increase their speeds, they get their actual speeds correct. Comcast has done the same thing, I had their 6Mbps plan at one point, I got 2.5Mbps usually and sometimes 3Mbps, so they're all doing the same thing. In closing, with all these speed increases, why is my Internet not getting faster?" What practices and tools do you use to test your bandwidth speed and how have you approached your ISP when the performance repeatedly fell short of your expectations?

One thing to note is that you'll never get the top speed advertised for any connection due to transmission overhead; even so, you should be able to get close (within about 10-20%). Also, ISPs oversell their bandwidth, so if you run your speed tests when other customers are using their connection, you will notice the performance hit.

Shocking!

[fragmentate]

On cable connections, you're sharing the connection -- is it so shocking that you're not getting what's advertised?

Then there's the whole issue of the internet in general. We've seen sites that are probably paying for OC3s and DS3s for their sites and you go visit their site and there's bad latency.

Then I click on my Slashdot bookmark -- voila! The explanation, darn Slashdotters hogging up all the bandwidth.

The point being there's a lot of noise in between the last hop out of your ISP and the destination address. Get over it. It's not false advertising, it's the unpredictability of the internet. Trying doing speed tests [speakeasy.net] to many destinations.

What does gramma need with 3Mbps anyway?!

Guess I'm lucky

[kaufmanmoore]

I've lived in 4 different places in 2 cities and have been able to consistantly recieve 80-90% of the advertised speed through cable. There's a reason why the ads on TV have speeds will vary written in fine print.

Municipal Broadband

[mysqlrocks]

Do what the city I live in did and start your (the citizens) own ISP [burlingtontelecom.net]. I get the speed I pay for on a fiber optic connection. Plus they offer TV and telephone service. Better service, cheaper rates, and it's owned by the people that use it.

I suck up.

[patryn20]

To be blunt, where I am there is only one choice for internet service. The single provider may change, depending on what municipality, but in the end you only have one choice in your apartment. So, when I have an issue I suck up. I act stupid and helpless and ultra sickly sweet. I thank them profusely every step of the way.

It may not be as satisfying as being intelligent or righteously indignant on the phone, but it gets great results. I consistently get a tech out same day (from ATT (SBC), no less). I have problems where my circuit speed will drop drastically (from 3Mbps to 145Kbps) on a regular basis, and now that I have started being saccharine sweet, it is generally fixed almost immediately.

Simply point out that it is running incredibly slow (say something about images and pages taking FOREVER to load, don't sound techie) and that you logged in following THEIR instructions (thank you guys for giving me those previously, oh thank you thank you) and checked the speed and saw that it was slower than normal (from what you guys told me before), and that you would greatly appreciate it if they could fix it (since I am so helpless and LOVE you guys), and please help me, and oh lord thank you so much for giving me your time.

Other than that, make sure your router isn't causing you problems. Swap it out with a borrowed one or something. I had a bad one that was destroying my throughput. Check cables, wall sockets, everything. Make sure you can eliminate everything on your end before you call them.

However, if they ask you to test things again, gleefully (pretend) to do it. It makes them happy and gets you better service later. After all, it is not really that hard to sit there reading the newspaper and drinking coffee and simply saying "Nope, still doesn't work."

Some things I've found.

[jafo]

One "trick" they use is that in our area (Colorado, QWest), the DSL speed rates they quote are all the ATM frame rates. ATM has around 20% overhead, so this means that a 1.5mbps line will give you more around 1.25mbps throughput.

I don't recall that I've ever gotten anything less than that on DSL across the line. I've run routers handling the "megacentral", the ISP end of the DSL connection, and have had more than a bit of opportunity to test DSL connection performance.

As far as cable, we have Comcast in this area, and are paying for the higher service level. I do notice that when the school year starts, we tend to have performance issues for a month or two. This has happened on several occasions. So, instead of 6 to 8mbps (they recently upgraded to 8mbps, before that it was 6), we get more like 3 to 4. Annoying, but not a huge issue.

I have noticed that on the Comcast sales literature, they say "N mbps *" where the * links to something saying "No guarantees".

However, most of the time I'm able to get 8mbps, when the remote end can handle it. I have servers hosted at a location where I know I have plenty of bandwidth. I just downloaded the Ubuntu Dapper ISO over cable:

730740736 bytes transferred in 710 seconds (1005.4K/s)

So, that's right at 8mbps. This is not unusual.

It's important to realize that there are several places where there could be performance issues though. The line, the directly connected ISP bandwidth, the server you're downloading from, and everything in between.

Winging at your ISP for problems which are outside their control isn't going to be helping anyone. If you are downloading Dapper right now via FTP from the main site, the server is almost certainly not going to be able to handle 8mbps.

Another thing I'd wonder is whether maybe your grandmother might have a virus or two, or perhaps there's some file-sharing going on? All these lines have a fraction of the upstream bandwidth that they do down. If you are pushing out much data, it interferes with incoming data. If you do any performance testing, make SURE that you don't have anything else using it, either outgoing or incoming.

Hope this helps.

Sean

Speed Tests Reliable?

[fohat]

My experience in this 65 year old apartment building is that the copper wire here won't support DSL above 4 megabits. I recently switched to cable internet (Comcast, Maryland) and saw a huge increase in available bandwidth. They originally promised a higher upload speed (I apparently purchased 6/384 and thought I was getting 6/768) and when I called to inquire I was offered 8/768 for 10$ more a month.

I'm able to pretty much get full speed out of my connection, but most of the times when I do speed tests on say Speakeasy.net or through other test sites, I frequently get reports that indicate half of my potential speed. I have been wondering if perhaps these tests are not very accurate at all, and would suggest connecting to a nice fast torrent to get a feel for how fast your connection is.

Works for me, anyway.

Re:Shocking!

[taylortbb]

Around here it seems everyone bashes cable, but it's not the technology that sucks, but the providers. I don't know what available in your area but here (Toronto, Canada) I have the choice of Bell (DSL) or Rogers (Cable) for highspeed. On a 3 Mbps DSL line ($40/month CAD) you'll likely pull 3-4Mbps from any decent server. Over one of the 6Mbps cable lines ($46/month CAD) you can generally pull about 10Mbps or download speeds of 1.25MB/sec sustained.

My suggestion to the poster is try a different ISP, they're not all bad. I don't know your location so I can't be more specific but if your in the US you're sort of stuck with the fact that all the ISPs I've seen down there suck.

Re:Don't have that problem with my fiberoptic

[hivebrain]

I pay $55/month for 30 down / 5 up. I think you can get 15 and 2 for about $40. I also get my TV through the fiber now -- $30 cheaper than digital cable through Comcast, more channels, HDTV, yadda, yadda.
I'm in the northern VA suburbs of DC and I know that Verizon's already in a wide variety of towns in the area.

Re:SLA?

[Columcille]

Wow. From internet connection speeds to fears about impinged freedom. Does everything become political?

Re:Municipal Broadband

[mysqlrocks]

It's a cooperative?

No, it's municipal broadband as my subject line said. That means that the city runs it. We, the people, collectively own the assets of the city since we are the voters and the taxpayers and this is a democracy. Since they only provide service within the city, then everyone that is able to receive service owns the network. We "own" it in the same we that we own the parks and other public spaces.

If I don't like the way the network is run I can vote to change it. Now, you may argue that I can "vote with my dollar" if I am customer of MegaCorp Broadband. The problem with that logic is that not everyone has an equal vote. In a democracy, everyone gets one vote no matter how much money you have. We, the citizens, decided we were tired of getting screwed by MegaCorp Broaband (Adelphia or Verizon as the case may be here and now) and that we would have provide our own service. Now, I can get my Broadband, my electricity (yes, the electric company is run by the city here too), and my water from the city and I can feel confident that I, as a citizen, can have a say in how these services are run regardless of how much money I may have.

Re:No surprise here move along

[ScrewMaster]

AT&T was a past master of the art of the lobby, and being taken over by the likes of Edward Whitacre and SBC doesn't change that fact. Worse, Comcast, AOL, Verizon and the rest are just as much for this tiered crap as phone companies are. The rich keep getting richer ... this is just more of the same, I guess. While I fault Whitacre and his particular brand of corporate criminal for their misdeeds, I'm especially disappointed in Congress, who should know better.

too many ways to point the finger of blame

[davmoo]

Why are you so sure the problem is your ISP? Do you know for a fact that the speed test is accurate? Are you doing the speed tests during a time of peak internet usage? Are other sites that you are connecting to serving fast enough to fill your pipe at full speed? If you are connecting to a site that can only serve 1 mbps, I don't care how fast your speed is promised to be, you'll never get anything from that site faster than 1 mbps.

And be careful when making claims "no ISP delivers the speed they promise". My ISP is Comcast on a cable modem. They claim they are giving me 6 mbps. And 99 percent of the time when I'm doing big video or Linux iso downloads or such like that and can see a good test of my actual speed, I'm getting the speed they say they're selling me...6 mbps.

Re:Municipal Broadband

[yuvi]

Same thing where I live, although the co-op services quite a few areas across two states. I pay for a residential 1.5 Mbps / 256 Kbps cable connection, and consistently get the rated download speed. In fact, I've gotten up to a 2 Mbps download sometimes, and consistently get 360 Kbps upload. All thanks to a small local ISP (the big national ones don't even cover my area anyways, so it's my only option.)

Re:Municipal Broadband

[HardCase]

And none of this is funded with tax dollars collected from the general public?

The answer is in the link that the original poster provided. But since this is /., I know that nobody can be troubled to read the article. From the FAQ:

Are my taxes going into this project?

No.

I'm also getting my full bandwidth

[Lenolium]

Thank the bandwidth gods for UTOPIA, a community fiber-optic system. 15Mbit symmetric. I've had LAN's slower than this, and I get a 2ms ping time to XMission's border router. Logged on to counter-strike, and found a few games being hosted at my isp with under 10ms pings. It's amazing what can happen when you get the damn telcos out of the way. :::.. Download Stats ..:::
Connection is:: 14320 Kbps about 14.32 Mbps (tested with 12160 kB)
Download Speed is:: 1748 kB/s
Tested From:: http://testmy.net/ [testmy.net] (Server 2)
Test Time:: 2006/06/01 - 11:34pm
Bottom Line:: 250X faster than 56K 1MB Download in 0.59 sec
Tested from a 12160 kB file and took 6.956 seconds to complete
Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.0.3) Gecko/20060326 Firefox/1.5.0.3 (Debian-1.5.dfsg+1.5.0.3-1)
Diagnosis: Awesome! 20% + : 85.68 % faster than the average for host (xmission.com)
Validation Link:: http://testmy.net/stats/id-QIOGKAJMB [testmy.net]

Re:speakeasy for both

[sulam]

I have Speakeasy as well. I bought a 1.5M down, 768K up DSL line initially. Later they started selling 6M down, and I upgraded to that. My line was a little unstable, probably due to older wiring in the building, and we lowered the speed to 4Mbps in order to keep it stable. They were apologetic at not being able to give me the full 6Mbps, and knocked a portion off my monthly to compensate.

Every time I call the service is fast and efficient. Yesterday I needed another static IP assigned, and it took literally 60s to do with the first tech I got on the phone. The guy obviously knew his stuff, despite being tier 1 tech support for them.

Basically I can't recommend them highly enough. You get what you pay for.

Re:Your link is probably fine...

[Meest]

The only problem is when your ISP doesn't support your OS and they tell you your on your own. So then i have to dig out my laptop with a windows 2000 install just to do their script the follow.

Then they tell me 90ms in their network is "Within standards"

I live in North Dakota.

Normaly I Ping 35ms to chicago based servers.

When I'm pinging 250ms to the same servers I do know that something is wrong. that is unacceptable to me.

This is what i would tell my ISP every time I Called them. They would say sorry, but their was nothing they could do. I told them then that I would just be calling back every 30 minutes until the issue was resolved... Here's the thing. STICK WITH IT!

I called my ISP every night about 5 times for 2 weeks strait. I had 4 technician's come out to my house in 3 weeks for service calls. After all that they finnaly realized i meant business. They split my node and now I'm all happy.

You just have to let them know that you won't let them walk over you.

Shouldn't that be standard procedure?

[Anonymous Coward]

Sorry, I'm not in the US. I live in Germany, and I know that *each and every* DSL provider here will send you a free DSL splitter with your contract. I have never heard of any other solution before reading your post :)

I have a DSL 6000/660 plan, my line is currently synced to 5120/736, which translates to 4637/667 after protocol overhead. I think that's fair enough, given the options of paying 3 EUR less for 3000/384 or 10 EUR more for ADSL2+ 16000/800. I also noticed that the backbone connection is very good - I often get 500kb/sec download rates and 74kb/sec upload.

My ISP undersell...!?

[grahamsz]

I ditched comcast for a local fixed wireless ISP (Mesa Networks) who seem to be holding customers despite having both DSL and Cable in the area.

I'm paying for a 3Mb/1Mb connection, yet according to the speedtest on speakeasy's site i'm actually getting 4022kbps/1044kbps.

If I use more distant speed test locations then it seems to be closer to what i'm paying for, however it looks like they must have raised the cap on the local end so that I can get transfers at the speed i'm paying for. On top of that, my connection bursts to 9/3 which makes small transfers really snappy :)

Re:Bit Versus Byte

[hyfe]

I know, a byte is 8 bits, but as a rule of thumb, dividing by 10 seems to include overhead.

Overhead in converting from bits to bytes? :)

Application-level measurement of bandwidth is of course actual data free of padding (since the padding is done by a lower networklayer, it's completely transparent).

That's not slow

[Jafar00]

I pay for a 20Mbps connection, yet I get barely 1.5-2Mbps. That's the thing that sux about living in a poor area of town. Nobody can be bothered fixing up the local telephone exchange. :/

Re:False advertising != lack of SLA

[Detritus]

Now, there are certain exceptions. In general, you can't drive a dense network at much beyond 1/3 the rated speed - thin-wire ethernet was bad for that - so you can expect similar sorts of problems on a shared line such as cable. The entire design of cable - a single line with taps off it - is exactly what thick-wire and thin-wire ethernet were like.

Broadband data networks over CATV are very different than shared-media Ethernet. Ethernet uses baseband signalling, everyone shares a common channel (CSMA). With cable, there can be multiple independent downlink channels. There is a single uplink channel that uses TDM to support multiple users. Each cable modem is assigned a shared 6 MHz downlink channel and a time slot on the uplink channel. There is no contention for access to the media.

Re:No surprise here move along

[rmerry72]

Is if fraud when hard drive companies sell you a "250GB" HDD?

Well, no, actually. A 250MB hard drive is exactly that: 250,000,000,000 bytes. That's the same definition of 1GB (ie 1,000,000,000 bytes) that ALL hard drive manufacturers use, and have been for quite a while. Most will actually state x,000,000,000 bytes. So its fair, and you're are getting what you pay for. It's the same thing here, you pick the description that makes you look the best.

Only if that description is acturate; Otherwise its fraud. Marketting is about putting the best spin on what you've got to sell. Saying your selling something your not is fraud. And marketing in general strays very close or a touch over the line.

With bandwidith it is reasonable to expect some loss, but you should be able to get at least 90% most times, and then deduct network and traffic overheads. If your only getting 25% of the bandwidth delivered that's a cause for suspictions of being ripped off.

Overselling is a poor excuse, one used by airlines all the time. If I buy a seat on a flight I expect to get one - even if everybody else they have sold a seat to does as well.

Re:No surprise here move along

[The_Mr_Flibble]

Well I don't know what it's like over in the states but over here (uk) there's a little something the providers don't make a large fuss about and that's the contention ratio
and in some cases it's really bad (something like 40:1 on a standard home dsl) so if you have a 2 meg adsl so if you only get connection speeds of 100kbs there isn't much the provider will do about it (yes I have seen this happen with a certain large provider over here who would not do anything because it was indeed in range)

So have a look at the contract you signed for your dsl and see what sort of ratio they try to give you.

If you have a problem with contention ratio then you may need to pay a bit more for say business broadband which has a lot lower contention ratio.

btw over here if say you have a 1mb adsl and you pay £20 a month for it and all your traffic goes over transit connections (which isps have to pay per mb for) and you use that 1 mb constantly the isp will loose a lot of cash as they will be paying about £30 a meg for the transit then they still have to pay the cost of the dsl connection as well.

In conclusion read the contract!
 

my cable provider delivers

[Cederic]

I have a 10Mbps cable connection. Sure, most 'net servers aren't able to give out files that fast. But the ones that are..

3-4 weeks ago I downloaded a 142MB file. Firefox reported it as coming down at one megabyte/sec. I'm not sure whether it lied, but the file was downloaded in under 2 minutes.

Surprised the hell out of me. Made me happy.

Cable company is NTL. Their technical support is absolutely atrocious. Luckily their connection is very stable, so I rarely have to call them. And the download speed is very nice indeed.

Re:No surprise here move along

[keraneuology]

After the promotional period Comcast said they were doubling my rate to just under $65/month. I spoke with three different 800-operators and sent an email to corporate asking them to reconsider and they refused. A couple of them told me outright that if I really wanted to switch to DSL ($17.99/month vs $6x.00/month) I should go for it, but that they do not consider price reductions under any circumstances.

Bye bye, Comcast.

Re:SLA?

[mo^]

I must admit to always thinking that the Car Analogy should be included as a form of Godwin's Law.

Re:my dsl, my test...

[David Horn]

Have ntl cable in the UK, looks like they have a different business model from the American companies as although the connection is supposed to be a fairly sluggish 10mbits, I just did a speedtest and got 12.5. :)

Re:SLA?

[tinkerghost]

To be fair, you ISP only owns parts of the highway: the on ramp (some might own a bit more than that). I bet you'd find that a bandwidth test against a server in their network would probably report numbers very close to what they're selling you. But there are quite a few bottleneck on the internet, including the bandwidth connection of the website you're trying to reach (including the bandwidth test sites I have seen).
I might buy that for the overall issue of not recieving the bandwidth promised. However when you concider:
She downgraded to the 768K plan expecting to still have 750K. Wrong, instead her speed dropped to 300K.
Note that all the numbers are in bits per second since he referenced them that way earlier in his statement.
You can see that the problem is not a bottleneck issue. If your 3Mb/s connection generates 750Kb/s and the problem is a bottleneck, then dropping the maximum speed available to you is not going to change anything. Your throughput at the bottleneck will be just as fast - 750Kb/s.
This is more likely a QoS implimentation which assigns specific allotments of bandwidth to the various levels of service. "OK, we have 100Tb/s of bandwidth, our 3Mb/s customers pay the most so we will give them 50Tb/s, 2Mb/s gets 25Tb/s ... Oh, and that sucker paying $20 a month for his dialup ... ehh, cut him off & tell him it's his modem is broken."
You can see the difference between bottlenecking & segregated bandwidth issues. If there's a bottleneck, everyone up to the throughput of the bottleneck doesn't know it's there. Everyone over that limit sees it exactly the same. With the bandwidth segregation, each tier will show differently based on the load at the time.

My strategy

[bradleyland]

When doing DSL installs for customers, my strategy is always to place the DSL modem closest to the point where the telco wiring terminates on the side of the house. I then use ethernet, powerline, or wireless to get the signal to the computers. Also, I've noticed that some DSL modems are much, much more sensitive to line conditions than others. For example, in Florida, Bellsouth offers a base Westel 6xxx series modem, a Versalink 4-port ethernet with WiFi DSL modem/router, and a Netopia 4-port ethernet with WiFi DSL modem/router. The new Netopia seems to be the most resilient in terms of connection drops and line quality tolerance.

I also picked up a SunSet DSL test-set on eBay for just under $100. This is the same equipment that Bellsouth uses to test the line. I've found it to be very, very useful. I unplug all the phones in the house and test jacks until I find one that produces no errors.

Re:Bandwidth Speed Tests

[dch24]

I really should use firefox, but I'm on a terminal only right now.