h a l f b a k e r y
You think: Aha! We go: ha, ha.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
On every Wi-Fi capable device I've used, there is a network description screen available which tells you things about the various networks it detects, such as the SSID, security type, and speed. And the speed they list is ALWAYS wrong. I've never understood where they get this particular metric, but
doesn't make sense from an end-user perspective. For instance, I'm currently sitting in a Fred Meyer grocery store, and the speed listed for the complimentary Wi-Fi network I'm connected to is '54 Mbps'. I've run a variety of speed tests, using apps and websites, with and without other users sharing the network, and the highest I've managed to pull on any of my devices is 2 Mbps download, 1.5 Mbps upload.
So I don't know what that speed metric that says '54 Mbps' is supposed to mean (so I'm NOT suggesting getting rid of it, for those of you who may know what it means and are able to use that information) but I'd like an average user speed, the speed the average user actually gets, listed in the network analysis data so when I'm trying to download a large file, I can 'shop around', so to speak, for the fastest network before I start the download.
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
||Listen to Mr. I Can't Get Enough Bandwidth in the Grocery
Store complaining. [Qwest], I have to walk down my
stand next to my mailbox just to get enough bars to check
||[Alterother]: Complain to your cell provider. Loudly and
repeatedly. If you have an iPhone on AT&T, there's an app
that allows you to file reports of places you get no service;
there may be similar apps for other phones/providers. My
parents had lousy coverage at their house, and they
complained to AT&T enough times that they eventually
were sent two picocell units free of charge.
||I don't have an iPhone, I don't even have a smart phone,
and my cell provider will politely tell me that if I'd just
kindly move a little closer to civilization, they'll be happy
to provide me with all the coverage I need, but as it stands
they would literally have to move mountains to get a
better signal to my house.
|| My point is that not everyone has access to wifi
everywhere they go, and I frequently hear those who are
accustomed to it complaining about the poor quality of
what is, essentially, a privilege. I'm sorry if I come off a
little snide, but I grew up in a house where, if I loudly
made it known that I didn't appreciate something nice that
had been given to me, it got taken away.
||I'm not complaining about the speed, Alter, I'm simply asking for
the speed to be accurately and clearly stated on the already-
existing network analysis page.
|| Ytk, what you refer to is called a femtocell. AT&T's version is
called the Microcell, and they VERY rarely give them out for free,
no matter how loudly and frequently you complain. I've worked
in their tech support department for 3 years. They're also only
available, even for purchase, in certain areas. Even if you buy
one cheap off eBay or Craigslist, it won't work if the GPS unit
inside the Microcell detects that it's not within an authorized
coverage area. You can thank the FCC for that.
|| Complaining about the cost usually gets you a referral to the
EULA, which basically says there may be scenarios where
additional hardware is required to provide service in certain
places, and the customer bears responsibility for the cost of
obtaining such equipment.
||//they VERY rarely give them out for free//
|| Not anymore. As best I can tell, once they lost their
monopoly on the iPhone AT&T started handing them out
party favors once it was clear that people were jumping
ship in droves based on their mediocre coverage.
|| [Alter], it doesn't hurt to try and complain. It's not a very
expensive device, less than $200, and it basically routes
your cell phone connection seamlessly over your broadband
line. My parents were going to buy one because of their
consistently poor coverage, when I pointed out that they
shouldn't have to PAY AT&T for the privilege of extending
their cell network via their own Internet connection. It
really didn't take much complaining at allmaybe an email
or twoand they got an apology letter from some vice
president along with a coupon for a free microcell. One
unit wasn't enough to cover their entire house, so they
complained again, and received the same form letter along
with another coupon. Cell number portability has greatly
increased competition between providers, and since the
costs are largely similar between the major carriers they
have to compete on coverage, and they know it. If you
send them an email or two saying you're generally happy
with your service but frustrated about your lack of
coverage at home, and maybe mention that your friend
who uses carrier X seemed to get a better signal when he
came to visit, you might be surprised at their response. It
doesn't cost them very much relative to the revenue
gained by keeping you as a subscriber.
||I always thought they should be free with contract. Sprint gives
out their Air Rave femtocells for free. But whatcha think of the
idear? By the way, I'm now using a neighbor's Wi-Fi. The 'link
speed' listed is 65 Mbps. I just ran a series of speed tests,
pulling between 12 and 23 Mbps, with an average of about 16
Mbps. Mind you, that's more than
fast enough for what I need, but again... I'm not getting
anywhere near 65 Mbps, so why does that number appear?
|| Also, regarding Alterother's attitude that Wi-Fi is a privilege: Not
when it's being advertised as part of your data plan, it's not.
Most carriers are no longer offering unlimited data plans for
smartphones. AT&T says it's limited data plans 'include' free
unlimited access to over 20,000 hotspot locations nationwide.
T-Mobile uses the same hook with their limited data plans and
Wi-Fi hotspots. If I'm paying for a 3G or 4G data plan and Wi-Fi
is supposed to be an extension of that plan, then I should be
able to expect Wi-Fi speeds comparable to or exceeding
cellular network speeds, and that's not happening. Most AT&T
hotspots (you'll see them mostly at McDonald's, Starbucks, and
Barnes & Noble bookstores) provide speeds that are nowhere
near what I get on 3G, let alone 4G. T-Mobile's Wi-Fi isn't any
||The indicated speed of the connection is a simple mathematical equation of the total number of packets of data that could be delivered between your router and the server in a second. The mathematics behind this are a relatively simple, but could be summed up as representing the maximum frequency with which packets will be sent multiplied by the size of those packets. Packets are inherently verbose, especially when you consider error control and routing. When combined with the fact that every machine between you and the server you are retrieving data from gives your request progressively lower priority, often shuffling your packet back in time to avoid peak loading, etc, fragmentation, packet loss, suddenly it takes a long time to get data from a remote server, even though that data comes to the router in brief bursts of 56Mbps (which in reality it rarely does, because, like your router spreads the bursts that it receives over an averaging window to reduce errors, the routers upstream are also windowing data to the server, etc etc.) So when it all comes down to it, your router might have negotiated a peak rate with your host, your host might have negotiated a peak rate to the trunk, trunk to hub, hub to hub, hub to trunk and trunk to server on the other end, the data has to make the journey with all the other data on the network and thus get queried, requested, addressed,cued, packet sorted, bundled, cued again, and at the end of that while each leg of the journey was made at a speed higher than 56Mbps from the perspective of packet frequency, the final time to receive each packet is much longer. Does this help?
||Yes, it does... that clears up the question of what that number
means, thank you. What do you think of my idea for adding the
actual expected user-end speed to the analysis screen?
||A few years ago, my city made headlines in a few nerdpapers for having the highest average to-the-customer bandwidth in the USA. I did some checking around, and found that the "average" they quoted was higher (factor of 3) that the highest maximum bandwidth you could buy anywhere in this market regardless of price.
|| The publications said a retraction or correction would "not be feasable".
||I meant that wi-fi is a privelege in the sense that it is a
frivolous luxury that didn't even exist a few years ago, yet
now many people take it for granted and can't seem to do
without, and bitch when they don't have it. I see people up
at the mountain, sitting around in the lodge (which has
excellent wi-fi) bitching that they can't upload the 5-
minute video of their awesome ski run to yootoob or
f@cebook or whatever. These people came here and paid a
lot of money to ski at the top resort in the East, and yet
they're down at the base on the most beautiful day of the
year fucking around with their smartphones!
|| So, I'm sorry if it's a pet peeve of mine.
||I guess the main number that's missing from the metric posted would be the dynamic load on the router, i.e. the number of current users. Once you know how many people are sharing your wifi link at any particular time, you need to divide the total bandwidth by that number. If there are 20 people all busily wifiing away, the chances are that the available bandwith will need to be shared 20 ways, if the posted "top" speed is 20Mbps, that's only one Mbps each. Of course, it's not always as easy as that in practice, but it should give a better overall impression of the experience you can likely expect.
|| Back in the old days, the internet used to slow down around 2pm, just when all the Americans started logging on.
||Ummm, wait a second. You're aware that regardless of the
speed of the wifi connection, your measured download
speed will be limited by the speed of the Internet
connection, right? If so, then I have no idea what you're
talking about with this idea.
||Well like I said, I don't know or particularly care what the
existing 'link speed' metric means. I want a number on
the analysis page that tells me how fast my internet
speed will be. I want to know at what Mbps or Kbps rate
I can expect a photo of my dog to go from my phone to
my wall on Facebook after I press 'Upload'. I want to
know how fast I can expect a 1.5 hour debate video
from intelligencesquaredus.org to download onto my
device. If the bandwidth provided by the ISP to the router is,
say, 45 Mbps, but the router is a cheap Netgear N-100 router
bought at Wal-Mart for 20 bucks and only has a maximum
throughput of, say, 20 Mbps, then I should expect, with no other
users sharing the bandwidth, to get a speed of 20 Mbps. If the
same ISP provides a maximum bandwidth of 6 Mbps (like with
Comcast's 'elite' package) with that same 20 Mbps router, then I
can expect, approximately, a 6 Mbps user experience.
|| So I guess the metric I'm looking for would take the ISP speed
OR the router output speed (whichever is lower) and divide it by
the number of users on the network.
||Wi-Fi (802.11) runs at a maximum of 54 Mbps,
though the higher bandwidth variants of 802.11N
are starting to exceed that as a theoretical
|| However, that does not mean that you'll get that
speed to the internet, especially if you're using a
DSL connection. You might get 54Mbps to the
DSL router but the gateway speed of the router is
going to be somewhere around 1-20 Mbps,
depending upon the data plan assigned by your ISP
and the dynamic load on the shared DSL channel
(ISP port loading). I have a dedicated channel
from my satellite provider (too far out of town for
ADSL or reliable wireless internet) that no longer
falls in a hole every afternoon when all of the
neighbourhood kids come home from school.
|| You could get 54 Mbps internal transfer speed from
one LAN node to another, on a good day, if they
were the only two computers in the network.
When the number of computers sharing the Wi-Fi
connection goes up then the access speed goes
||// [Alterother] I'm beginning to think you run a rescue
shelter for peeves. //
||So basically the idea is the result you get from speedtest.net, just listed on the wifi settings screen so that you know before you connect?
||In a nutshell, yes, that's the big idea. Not a huge deal, I readily
admit, but it'd be awfully convenient for when folks want to
stream video or upload 50 recent pics to Facebook.
||Not that I begrudge a fellow Baker his/her right to cast a negative vote, it would be nice if such a vote were followed by an explanation of sorts.
||Posted in (personality) disorder.
||Sorry, my browser has been experiencing technical difficulties. I've switched to another until the situation is resolved.
||I find that the browser on Android 2.3.6
"GINGERBREAD" does bad shit to HB posts and
||I'm using a third party browser called Dolphin HD, the same I
always use. Been acting weird the last few days.
||I think it does it on porpoise, just to show you up.
||Oh I have no doubt. You'd never see that sort of higher-level
thinking on an iPhone.
||They're called iPhone, with a lower case I, because
the average user isn't that proficient at punctuation
|| It's the difference between knowing your shit and
knowing you're shit.
||Its a good idea, but wouldn't constantly sending packages to and from a speed testing website slow down your grocery facebooking experience even more?
||Needn't be constant. Run the speed test periodically, say every 4
hours. That's why I titled it 'average... speed'.
||I thought it was the speed that an average user gets?
||When I were a lahd, weren't no sooch thing as no
inter-net, so point's all moot, ah'd say. Be grateful
fer what's yer got, sonny.