how-to block ads
|
cracker 52
@comcast.net
| reply to dscline
Re: [News] PowerBoost
Right on dscline. That's what I've been trying to convey as to why my Speakeasy test results have been reasonably accurate and consistent. People in markets with prolonged or significant bursts (not Powerboost) think it's that way in all markets. But we certainly don't have that here in Atlanta. What occasional bursting that may occur will only peak at 10 megs for a second at best. When we finally do get Powerboost, then our test results will probably be skewed. | |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
 ·Comcast
3 edits | And what *I* try to convey is that because one market is getting semi-accurate speed tests, that it is not the same everywhere.
For *MOST* of the Comcast markets, speed tests are inaccurate.
Have a look: »/archive?zip=&···t=Search
I see alot of scatered ZIP codes there... and alot of rather high results.
I'll say it again. Speed tests are not to be trusted - any of them!
When you run a test, and it shows below normal speeds, does that mean your connection is hosed? Likely the test hiccupped - not to be trusted. 
Downloading a large file from a known fast server is a more accurate test of your line's capabilities.
Where ever you are.
Whatever ISP you have.
Whatever market you are in.
Just as when people state that Testmy.net is an accurate test...
:::.. Download Stats ..:::
Download Connection is:: 6195 Kbps about 6.2 Mbps (tested with 5983 kB)
Download Speed is:: 756 kB/s
Tested From:: »testmy.net/ (Server 1)
Test Time:: 2006/07/01 - 12:18pm
Bottom Line:: 108X faster than 56K 1MB Download in 1.35 sec
Tested from a 5983 kB file and took 7.912 seconds to complete
Download Diagnosis:: Looks Great : 19.73 % faster than the average for host (comcast.net)
D-Validation Link:: »testmy.net/stats/id-OWUYJ697I
User Agent:: Opera/9.00 (Windows NT 5.1; U; en) [!]
I don't think so! Sorry! LOL!
--
Think outside the Fox... Opera | |
Bens Delicious Hog
@comcast.net
 from:
dadkins 
|
Correctamundo dadkins. Only real way to judge speed is by observing a larege download. Even then it is dependant on the server providing the data and the route the data takes. Lots of folks don't seem to want to accept this.
The quick and easy speed tests are really just fluff. | |
cracker 52
@comcast.net
| reply to dadkins
Where did you get that any of us made any generalization that most of Comcast markets don't have bursting or don't have skewed test results? I only referred to Atlanta has not having much bursts and usually not having skewed test results. Just enter 303 in the archive. Unlike you, I'm not the one making generalizations.
I agree that the true test is from a FTP download. I just hate these silly sweeping generalizations, though the worst ones are from those who are having connection/speed problems.
Also, from the link you gave, the skewed results are not in any of the southeastern states, Texas, southern California or the northwestern states. Those aren't exactly small markets.
Download Speed: 8483 kbps (1060.4 KB/sec transfer rate)
Upload Speed: 721 kbps (90.1 KB/sec transfer rate)
I get those readings time after time. | |
danny9
Go Ahead, Make My Day
Premium
join:2002-07-14
Clinton Township, MI
clubs: 
 ·VoicePulse
·Comcast
| reply to dadkins
I agree. Most speed tests I have used show no consistancy.
The two I have found too be the most consistant are:
»speedtest.umflint.edu/ and »speedtest.mi.charter.com/
Has anyone used these before and have any opinions on them?
--
To Think or not to Think: That is the real question. VoicePulse 07/29/04 | |
Johkal
Cool Cat
Premium,MVM
join:2002-11-13
Happy Valley
clubs:
1 edit | It's all subjective, but I have always liked the Web100 sites. | |
dscline
join:2001-09-01
Atlanta, GA
| reply to dadkins
said by dadkins :And what *I* try to convey is that because one market is getting semi-accurate speed tests, that it is not the same everywhere. ... When you run a test, and it shows below normal speeds, does that mean your connection is hosed? Likely the test hiccupped - not to be trusted. Well that's completely different issue than netpersec showing significantly different results than the speed test DURING the speed test.
A speedtest is NOT a complex thing. It doesn't take a whole lot to automate the process of clearing the cache or use random data or whatever, then transfer it, and time it. I don't understand what about that process has to be inherently inaccurate. It should be accurate . Obviously, if you've got significant bursting/boosting in your area, it won't accurately represent the average speed of downloads of different sizes, just like a large ftp transfer won't accurately represent the transfer speed of a small file. It's not so much an accuracy issue, it's a sample size issue that manifests itself in a variable speed environment, which is what those of you in bursted/boosted essentially are (more so than the net already is).
But the puzzlement to me is why a speedtest would give significantly different results than the average while monitoring the transfer with netpersec, while some of us using the same test sites are getting very accurate results. | |
doomed999
join:2004-02-25
Romeoville, IL
1 edit | 
Screen Of Traffic threw Firewall... |
2006-07-02 09:58:50 Speed test @ speed.nap.wideopenwest.com 23806/358 kbps
2006-07-02 09:56:39 Speed test @ chicago.speedtest.frontiernet.net 15031/359 kbps
Explain how the speedtest is way up as the MAX though the firewall was 284k??
Just proof that as some markets the tests may be accurate, it sure isn't by me... | |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
·Comcast
| reply to dscline
Ok, say you have a 1 second pause(it happens) while doing a speed test. This will give the tester a low result.
Because the tester sees this low result, should they immediately call Comcast and complain?
Bursting.
When I see 14 - 46mbps results... should I think I have a magical line and then complain when an actual download shows the 8.5mbps I normally get?
Since Comcast has had bursting for nearly a year, I suspect that your market may have a problem.
Whether it is oversold, equipment that is overloaded/old, whatever... you get reasonably close results.
This does not mean that the Speakeasy test(or any other speed test) is to be held as accurate.
Yeah, NetPerSec shows an instantanious burst here of about 19mbps then drops... yet the Speakeasy(S.F.) test keeps on climbing to unreal results!
Santa Cruz NDT also shows me above 9.2mbps and that is running for 10 seconds.
I'm here in California. We get any and all upgrades/enhancements last!
I have been getting 12-14mbps for about a year on various different tests.
Some others, I see ~4mbps. Which one(s) should I go by?
I'll tell you which one(s) I should trust - none of them!
Seeing as most every other market is seeing odd results from speed tests... they all suck!
Ever consider that maybe, just maybe, the Atlanta Speakeasy test is congested?
Perhaps the Atlanta Comcast market is off a little?
People in nearly all other markets are seeing insane burst speeds...
Try this, go to the Speakeasy test and choose San Francisco, or any other Speakeasy server. What are your results?
--
Think outside the Fox... Opera | |
CajunTek
Insane Cajun
Premium,MVM
join:2003-08-08
Arlington, TX
·RoadRunner Cable
| Oh yeah they are accurate  | |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
·Comcast
| 
About right... |
Here, a real speed test! | |
dscline
join:2001-09-01
Atlanta, GA
1 edit | reply to dadkins
quote:
Ok, say you have a 1 second pause(it happens) while doing a speed test. This will give the tester a low result.
Because the tester sees this low result, should they immediately call Comcast and complain?
Bursting.
When I see 14 - 46mbps results... should I think I have a magical line and then complain when an actual download shows the 8.5mbps I normally get?
Of course not. And the same would be said if those results were based on large FTP transfers. There are no guarrantees, actual transfer rate can and does vary, especially if you have bursting/boosting, and you're downloading files of various sizes. The only time you should be upset is if you consistently get speeds significantly lower than what you're paying for. I fail to see what any of that has to do with discrepancies between speed tests and ACTUAL transfer rate during that time period.
quote:
Since Comcast has had bursting for nearly a year, I suspect that your market may have a problem.
Whether it is oversold, equipment that is overloaded/old, whatever... you get reasonably close results.
Keep in mind we're firmly entrenched in Bellsouth territory. There's not much motivation here for Comcast to improve our bandwidth, and with BS's luck in controlling their competition (outside of cable, which itself is inherently monopolised), I don't see that changing any time soon. Does that mean there's a problem in our market? Guess that depends on your definition. I think there's definitely a problem with the competitiveness in our market, but not in the equipment. While I don't get the bursting many get, I do have very consistent service. It's extremely rare that I get performance less than what it should be for my 8mb tier that can't be attributable to the source I'm downloading from.
quote:
This does not mean that the Speakeasy test(or any other speed test) is to be held as accurate.
I'm not saying they are, or aren't. But they've certainly appeared to be very accurate for me. That doesn't mean they are for other areas. In fact, if you live in an area with bursting or boosting, I would EXPECT the results to be higher than averages from longer downloads. What I still don't see is an explanation as to why the actual measured throughput by netpersec is significantly different than your speedtest. Our results should be different, because the speed characteristics are different in our two markets. But the CODE that the speedtests use should be the same if we're using the same speedtests, so I don't understand why yours are screwed up, and mine appear to be on target.
quote:
Yeah, NetPerSec shows an instantanious burst here of about 19mbps then drops... yet the Speakeasy(S.F.) test keeps on climbing to unreal results!
And that is what I'm curious about. The only logical explanation I can think of is that the tests are written to look at the peaks, assuming that slower spots are due to net congestion. Since my bandwidth is fairly constant, this wouldn't result in different results for me anyway. But certainly would for someone with bursting/boosting. Do your speedtest results correspond with the PEAKS in netpersec?
quote:
I have been getting 12-14mbps for about a year on various different tests.
Some others, I see ~4mbps. Which one(s) should I go by?
I'll tell you which one(s) I should trust - none of them!
And I get significantly different results with large ftp downloads from various sites. Some are consistently fast, some are consistently slow. Some vary. So *I* believe the ones that are consistently fast, because it's logical that the many variables between me and the slower and/or less consistent servers can hinder the results. Since I don't appear to have any significant boosting/bursting, there aren't any variables I can think of that would affect my results positively beyond my caps. And the results have proven that. If you've got significant bursting, then your bandwidth is essentially variable. It seems the problem is people in those areas are expecting consistent results, and they just won't be.
quote:
Perhaps the Atlanta Comcast market is off a little?
People in nearly all other markets are seeing insane burst speeds... I'm here in California. We get any and all upgrades/enhancements last!
Ok, so which is it? Your market sucks more, or mine?
quote:
Try this, go to the Speakeasy test and choose San Francisco, or any other Speakeasy server. What are your results?
They are above (don't know how to make them show up at the bottom). SanFran, Atlanta, and OOL, respectively. The Atlanta result is consistent with what I typically get (between 7900 & 8200). The upload is usually a little higher when I have my QoS turned off (I use VoIP). I don't do QoS on the download side, as I've not found it to be needed. I don't normally test SanFran, but not surprised to see it lower, being on the opposite side of the country. The OOL test is consistent with other sites that have high bandwidth, but of course, in real world use, most places I visit don't have as much bandwidth.
| |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
·Comcast
1 edit | 
Atlanta | 
San Francisco |
As you can see, your market is apparently experiencing congestion or is a year behind on Comcast's Burst implementation.
This does not rule out speed tests being unreliable.
Gathering that my entire city was re-fibered(HFC laid late 2004), I would say that I have the better connection overall.
Your area/market may still need work, sorry! 
OOL, here, is slow as mollasses. It's a routing thing I imagine.
If you look at my previous post, you will see that a Knoppix torrent is rather stable. THAT is a Real World download in progress.
Speed tests are not to be trusted, I don't care where you are, or what market you are in. Even if it does show relatively close results, it uses too small of a file to accurately gague the line's speed.
See pics.
I can get nearly 8000k from the Atlanta Speakeasy... what does that tell you?
NOTE: I'm in Cali.
The S.F. test is ~14 miles from here, naturally I have a better connection to it, as opposed to ATL which is a few more miles away.
You make the call.
--
Think outside the Fox... Opera | |
motoracer
join:2003-09-15
Valencia, CA
1 edit | Dadkins,
Can you stop preaching about how inaccurate speedtests are? Everyone, I mean everyone, knows your view of them, so just drop it once and for all please. Thanks  | |
dscline
join:2001-09-01
Atlanta, GA
| reply to dadkins
said by dadkins :Thank you! As you can see, your market is apparently experiencing congestion or is a year behind on Comcast's Burst implementation. As I've said all along, we don't appear to have significant bursting here. But I don't follow your logic that the results show that my market is experiencing congestion... tests to a local speed test server are consistently fast. Tests to a known fast server are consistently fast. Tests to a far off speed test server are inconsistent and slow. How in the world does that point to local congestion? You get fast results to the ATL server, faster results to SF, slow results to OOL, but in YOUR case the slow OOL is a routing issue?  The only thing this seems to prove to me is that a) you get faster service on shorter downloads because you have bursting, and b) greater distance adds more variables that can slow bandwidth and increase variability. I don't think there's any surprises there.
quote:
This does not rule out speed tests being unreliable.
I don't see what it has to do with speed tests, period. It's more an example of the inconsistent nature of transfers between different sites and various distances, whether it's a speed test, ftp download, web page, or whatever.
quote:
Gathering that my entire city was re-fibered(HFC laid late 2004), I would say that I have the better connection overall.
Your area/market may still need work, sorry!
Ok, so I suck more. Does that mean I win or lose?
quote:
Speed tests are not to be trusted, I don't care where you are, or what market you are in. Even if it does show relatively close results, it uses too small of a file to accurately gague the line's speed.
Ok, so what I've seen you promoting in the past is to use large ftp downloads. But my large ftp downloads from reliable servers yield the same results as my local speakeasy tests. So are the ftp downloads also inaccurate?
Again, the whole issue has to do with sample size and bandwidth consistency. We don't have significant bursting here... my bandwidth is very consistent so the sample size doesn't play a big role in the results (oh wait, I have local congestion... I guess I'm consistently congested).
While this has all been fun, I'm still curious about the inconsistency you experience between speed tests and netpersec. | |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
·Comcast
| I give up! You're right, Speakeasy is accurate.
I am honestly sorry that you'r area is not getting what most other Comcast markets are.
--
Think outside the Fox... Opera | |
dscline
join:2001-09-01
Atlanta, GA
| said by dadkins :I give up! You're right, Speakeasy is accurate. First off, I never meant to insinuate that Speakeasy is accurate for you. Second, what you posted there doesn't show it to be accurate OR inaccurate. Now, if your netpersec showed a different result for the time sample of the download, THAT would be an indication of inaccuracy. You indicated in a previous post that that is indeed what you are experiencing, which is what I find curious, and am interested in the reasons why, but for some reason you seem satisfied to just accept that the discrepancy exists, and continue with your mantra that the inaccuracy exists for everyone.
quote:
I am honestly sorry that you'r area is not getting what most other Comcast markets are.
Eh, that's the breaks. But thanks for the sentiment. While it would be nice to have the bursting, and I'm sure it has some practical value, I'm not sure I'd really notice a big difference in my usage, outside of speedtests. As it is, it seems that my real world speed is more dictated by who I'm connecting to than my potential speed. As it is, I'm only paying for the 8MB tier for the upload increase. | |
cracker 52
@comcast.net
| reply to dscline
Actually here in Atlanta, several techs I spoke to said ATTBI spent about 2 million dollars upgrading the headend, network and plant just before being acquired by Comcast. The hybrid plant is either 750 Mhz or 860 Mhz. If you look at BBR Comcast TV board, most people are envious of the Atlanta market having the one of most extensive channel lineups in the country (just ask the knowledgeable people in New Jersey who have the smallest lineups).
In some of the denser areas, capacity could be close to straining but, as the case with dscline, I've been getting consistent, full Gold tier speeds day and night. My parents live in an older community part of Atlanta, where you won't find many power users, and they don't get any bursting. | |
dscline
join:2001-09-01
Atlanta, GA
| said by cracker 52 :
In some of the denser areas, capacity could be close to straining but, as the case with dscline, I've been getting consistent, full Gold tier speeds day and night. My parents live in an older community part of Atlanta, where you won't find many power users, and they don't get any bursting. When I first moved in to this house 13 years ago, this was somewhat the hood (near Little Five Points). I was really surprised when I got cable internet back around '98 before many of the areas that you'd consider were probably much bigger markets (back then, it was MediaOne, before it became AT&T, then Comcast). For the longest time, I was probably one of the few in my area to have broadband. In the past couple of years, this has become quite the trendy (and much more expensive) place to buy, and there are huge new lofts and other developments going up all over. I'm fully expecting the near spontaneous influx of likely broadband customers to catch the local infrastructure by surprise, but thankfully it hasn't appeared to happened yet. | |
dadkins
Can you do Blu?
Premium,MVM
join:2003-09-26
Hercules, CA
·Comcast
2 edits | reply to motoracer
Accurate? |
said by motoracer :Dadkins, Can you stop preaching about how inaccurate speedtests are? Everyone, I mean everyone, knows your view of them, so just drop it once and for all please. Thanks Uh huh... Oh yeah! They're accurate, huh? LMAO! 
For some odd reason, I seriously doubt that Comcast is hitting me with ~5 MegaBytes per second. 
No one is forcing you to read or reply to anything I (or anyone else for that matter) posts. Are they? 
--
Think outside the Fox... Opera | |
|
