Home Forums Networks & Cross-Platform Networking DSL d/l speed plummets!

I don't imagine that there'll be a definitive answer to this question, but I'll ask it anyhow...

I recently noticed a precipitous drop in my Verizon DSL d/l speed, and when it returned to normal after I took my AirPort Express Base Station out of the equation it occurred to me that the change may have resulted from my recent switch from 802.11n (802.11b/g compatible) to 802.11n only (5GHz).

Sure enough, reverting to b/g compatible restored my d/l speed to its former "lofty" number.

Sooo... Since I've repeatedly read that the 5GHz band generally has less interference that the 2.4GHz band, I'm wondering whether anybody can suggest what my roadblock may have been?

Thanks.

That is curious.

First I am surprised that switching from one band to the other or even from one protocol to another would have any noticeable effect on internet download speeds Unless you have a DSL that is delivering greater than say 40 to 50 Mbps download speeds either 802.11n or 802.11g would be still faster. (up to 50 Mbps for 802.11g and 80 to 150Mbps for 802.11n) That would imply that your Wi-Fi (formerly called Airport by Apple) LAN is pretty slow. What is your actual download rate on each band?

By and large and on the average the 5GHz band is less subject to RFI interference than the crowded 2.4GHz band. By the same token 5GHz is supposedly better at penetrating obstacles because of the shorter wavelength. 802.11n is usually less subject to adjacent channel interference as they do not overlap while the 802.11b/g channels do have some overlap. However, there is often a variance between theory and the real world. I would guess you have fallen into one of those situations.

The first thing I would do in your case is get something like AirRadar and check out the band, channel, signal strength, etc. of every Wi-Fi signal you are seeing in your location. Then I would try moving the Airport Express around a bit. Sometimes a move of only a few inches can make a huge difference in signal output. (Moving my Time Capsule six inches further up the wall made a 25% improvement in signal strength as seen at my iMac.) There are other parameters that you could change if you had an Airport Base Station, but they are not tweakable on the Airport Express.

I am going to be eager to see hows this one works out.

Thanks for your input and suggestions; because of the way things are set up around here it'll take a bit of doing to touch all bases, but I will get back to you.

Sorry for taking an inordinately long time to respond, but I was both having a composition block and waiting to see whether any other shoes would drop.

I'll try to make this coherent, but first, and this will give this bit of context, when I test my d/l speed using Speakeasy, Speedtest, or Verizon, what, exactly, am I measuring?

Is it the throughput to my modem, my AirPort Express Base Station, or my MacBook?

Cutting to the chase, then...

The diminished d/l speed to which I referred in my initial post was that shown by the aforementioned testing services, and I was never able to recreate the anomaly, i.e. 5.45Mbps rather than my expected 6.45Mbps, after it corrected itself...no matter how many times I switched bands/protocols or to which or in what order I switched.

My Base Station is a refurb, so I wonder...?

Reporting now on events that transpired after I switched back to 802.11 only (5GHz):

My Safari-reported d/l speed (I use the OS X 10.6.8 v 1.1 Combo as my benchamrk d/l.) remained diminished (as it had been during the period of my initially reported d/l speed issue) from 780Kbps to 660Kbps, and it did not return to normal until, after any number of weeks, it got up to 720Kbps, then, a few weeks later, 760Kbps, and, finally, after many weeks, 780Kbps (although intermittently).

Further, my signal strength occasionally dropped from around 100% to around 50-60% (always in the dead of night), and that was accompanied on three occasions (the third being that final dropped shoe) by my AirTunes becoming so choppy...unlistenable that I had to turn it off. (The noise level shown by AirRadar was always minimal.)

I note that my AirPort menu bar icon menu bar icon has never shown less that a full 4 bars regardless of what was going on otherwise.

I also note that there is a nobody within range of my AirPort card using the 5GHz band, so I can't report on relative anomalies.

And, in closing, I note that I switched to 802.11n (only) 2.4GHz after that last iTunes episode, and things have been pretty much as-expected since then (although not that a terribly long period of time has passed).

(Edit: I realize that this has been neither necessarily complete nor coherent, so) Please feel free to ask any questions you think may be pertinent, and maybe I'll have some documentation, info, or insight.

Thanks.

By the way... In considering all of this, please give some thought to whether I've possibly experienced something about which the FCC or some other agency ought to know. (I believe that my neighborhood has already sent one resident to Gitmo without his having joined the Marines first.)

You're measuring the end-to-end bandwidth between the testing service and whichever of your computers you're running the test on. The end-to-end bandwidth is limited by the slowest link along that path. That's usually your DSL speed, but it could be the speed of the testing service's internet connection (highly unlikely considering that such a service would need an extremely fast link to be at all useful), or it could be congestion along any link in the internet (which, like any highway, occasionally has traffic jams), or it could be the link between your computer and your router (highly unlikely, given the enormous speed difference between that link and typical DSL links). Except under unusual circumstances, it'll be your DSL bandwidth.

A drop from 6.45Mbps to 5.45Mbps is hardly "precipitous". That speed drop is so minor that I'd just shrug it off, saying "DSL seems a little slow today".

The speeds Safari reports are MBps (Mega Bytes per second), rather than the Mbps (Mega bits Per Second) that the testing services report. That's why they're smaller by a factor of 8 and change. (8 bits per byte, plus some overhead for framing. Depending on packet size, the ratio could be as high as 10 bits on the wire to transmit 1 byte of data. To get your 8.25:1 ratio, the site must be sending you very large packets.)

But in the general case Safari would be reporting the speed for a download from some site that may not have as fast a connection as would be typical for a speed testing service. The site may also be suffering congestion at the time you're doing your download. Actual download speeds will vary widely from site to site and from hour to hour.

I think you're agonizing over differences that are too slight to matter. Relax, and enjoy an internet connection that is more than twice as fast as mine. There's nothing in your data to suggest any problem with your base station, nor with interference from neighbors, nor anything the FCC is going to care about.

The short answer is "yes". You are testing:

• The speed of your computer
• The speed of the connection between your computer and your router/modem
• The speed of your router modem
• The speed of the connection between your router modem and the telco central office. (this is a function of the cable miles between your sight and the central office or any relays as well as the condition of the physical cable plant and every connection in the link.)
• The speed of the internet as seen by the network connection at the telco central office
• The speed and load on the testing server
• The speed of the DNS servers you are using
• The instantaneous load on the network
• etc.

Rarely, if ever, will you achieve 100% of the DSL speed you have contracted for. Depending on your local situation 80% may be the best you can expect.

Thanks for the comprehensive response.

> A drop from 6.45Mbps to 5.45Mbps is hardly "precipitous". That speed drop is so minor that I'd just shrug it off, saying "DSL seems a little slow today".

I'd be inclined to do the same, but this speed drop, which was most unusual in that my speed test results are very consistent, was a unique experience that continued for weeks and ended as suddenly as it began.

I dunno. :shrug:

> But in the general case Safari would be reporting the speed for a download from some site that may not have as fast a connection as would be typical for a speed testing service. The site may also be suffering congestion at the time you're doing your download. Actual download speeds will vary widely from site to site and from hour to hour.

That's why I use the biggest Apple d/l I can find as my benchmark...to flatten out those server, time of day, load, etc. issues.

> I think you're agonizing over differences that are too slight to matter. Relax, and enjoy an internet connection that is more than twice as fast as mine. There's nothing in your data to suggest any problem with your base station, nor with interference from neighbors, nor anything the FCC is going to care about.

That's reassuring, but there's still some accountant left in me that responds to seemingly anomalous numbers.

In closing, though, can you hazard a guess about what could slam the 5GHz band so hard that it destroys my AirTunes experience?

Excellent response; thanks!

> Rarely, if ever, will you achieve 100% of the DSL speed you have contracted for. Depending on your local situation 80% may be the best you can expect.

Well... My contract with Verizon is for 7Mb, and polling my modem tells me that my "Speed (Down)" is 7616 Kbits/sec, so I guess I'm doing OK with my (speed-testing services) reported roughly 6.5Mbits/sec?

But I'll repeat my question to ganbustein, i.e. can you hazard a guess as to what could slam the 5GHz band so hard that it destroys my AirTunes experience?

Guesses I might hazard:

• Interference from your neighbors' Wi-Fi.
• Interference from nearby non-WiFi equipment (microwave ovens and such-like).
• Physical interference in the form of pieces of metal blocking and/or reflecting your signal.
  • Think immobile objects, like metal in the frame of whatever building you're in.
  • Think semi-mobile objects, like metal desks and computers and other appliances.
  • Think mobile objects, like refrigerator doors or cars in the driveway.
• Remember that just because you don't see anything move doesn't mean nothing moved. A reflective surface you can't see isn't therefore non-reflective to radio waves.

Just to add to the mix, one of my iMacs is hardwired (Ethernet cable, not wi-fi) to my DSL/router and it experiences variable and noticeable speed changes. I shrug it off as "AT&T is slow today".

Point well-taken, Ira, but would you be equally blase about the 4 solid months of "DSL seems a little slow today" that I lived through?

And, on another note, how can you shrug off speed changes experienced by only one Mac out of (how) many as "AT&T is slow today?" If such were the case, wouldn't it be expected behavior for all your Macs?

Thanks for taking a stab, but...

• None of my neighbors networks that can be picked up by AirRadar is running in the 5GHz band.
• Interference from a microwave may be a possibility, but the only kitchen from which it could come is close enough that I'd expect frequent, rather than rare, interference. (Some day when I've got the time and she's home, I may ask my neighbor to crank up her oven as an experiment.)
• There's nothing in my environment that could either block or otherwise affect my signal that isn't a 100% present fact of life; nothing, pretty much including my deuced Mac(hina), ever moves or changes.

Oh, well... I guess this is just one of those head-scratchers that will be generating dandruff for a (long?) while.

I completely agree that more than 1 hour, let alone 4 months, is unacceptable.

And the behavior is actually on all the Macs that are wired in; I only referenced one as an example

> I completely agree that more than 1 hour, let alone 4 months, is unacceptable.

That whole episode just plain doesn't make any sense to me, so I'm going to keep on playing with it on the out chance that there's something to be discovered/learned.

> And the behavior is actually on all the Macs that are wired in; I only referenced one as an example

Thanks for the clarification.

Might this network have some machine connected via ethernet cable that (mistakenly) is running AirPort as well?

Good thought, but nope; the entire network is my deuced Mac(hina) and my AirPort Express Base Station.

My ethernet port has virtually never been used.