Why Don't Download Speeds Reach Advertised Limits?

[drmike]

Hi all,

For years, I've been using broadband internet connections. At home, I had Verizon DSL for a while, and now I have Time Warner cable. While both of these are 1.5 MB/sec connections, I've noticed that, with each, the fastest speeds I ever get when actually downloading a file is somewhere in the range of 180KB/sec. I'll sometimes test this by downloading a file from Adobe or Apple, or a similar site run by a major company with many servers, and I still get those speeds -- nothing ever close to a megabyte or a meg-and-a-half per second.

At work, where I have a T1 line, the actual download speeds are faster, but usually nothing over 250-300 KB/sec.

Why is this? I assume this is normal, as I see these kinds of speeds everywhere. Why don't actual download speeds meet the connection speeds?

 

[patonb]

Because they dont tell you that the advertised speed its at the absolute perfect setup.. brand new hardware and lines, no one else on, and a virgin sacrifice to the god of latency.

 

[Lexluethar]

Because those speeds are just that, 'limits.' Meaning that is the fastest they will give you with your current connection.

180kb is extremely slow though, think of that as like 3 times the speed of dialup - rediculously slow.

You shouldn't 'test' the speeds by downloading files from servers because those can be effected by distance and load on the server itself.

Go to this website and choose a city near you. Run two or three tests and that will give you an accurate measure of your upload and download speeds.

Speakeasy Speed Test

 

[drmike]

Thanks for the replies. I ran several speed tests using the link you provided, Lexluethar, and, as I suspected, the connection is between 1.4 and 1.5 MB/sec. The last one I ran clocked in at 1.437 MB/sec. This is on a brand new cable internet connection. When I had Verizon DSL, the connection/download speed was about the same ... usually somewhere between 1.3 and 1.5 MB/sec.

However, when I download a file -- any file, from any server -- my real-time, actual download speeds never exceed that 180 KB/sec rate I mentioned earlier. The reason I test my broadband speed by downloading files from servers (such as Adobe) is because I'm trying to get what I call my own "real world" download speeds -- so the actual speeds I get when I download files, whether they're programs (such as Adobe Reader), music files, or movie files from a service like iTunes.

Just this evening, I hard-wired my laptop straight into the modem and did one of these "real world" tests -- got about 168 KB/sec on average. That's a *big* difference from the results I get from the speed tests. I just can't figure out why this discrepancy exists, especially when the speed test give great results. And, as I mentioned, it's always been this way, whether with DSL or cable. Perhaps, as patonb says, the lines are too old. But I'm really curious why the tests give nearly 1.5 MB/sec but any download never really exceeds that 180 KB/sec average rate I usually get. (And, interestingly, the switch I made from DSL to cable didn't increase those "real world" speeds -- they're about the same.)

 

[farinha]

Well I don't think you have bad download speeds, I think you're mixing bits with bytes. Your line has, probably, 1.5mbps (mega bits per second) which is equal to 187.5 MB\s (mega bytes per second). Being the case your download speed is great, almost the theoretical value!

 

[oldskool]

Because those speeds are just that, 'limits.' Meaning that is the fastest they will give you with your current connection.

180kb is extremely slow though, think of that as like 3 times the speed of dialup - rediculously slow.

You shouldn't 'test' the speeds by downloading files from servers because those can be effected by distance and load on the server itself.

Go to this website and choose a city near you. Run two or three tests and that will give you an accurate measure of your upload and download speeds.

Speakeasy Speed Test

I actually have a "faster" connection according to Speakeasy than Speedtest.net

 

[patonb]

You need to remember, that the server upload speed and number of connections is the main factor in d/l actual stuff.

If they upload at 500, and 3 people are d/ling, you only can get 1/3rd of the 500...

 

[drmike]

Thanks for the replies, everyone. Now that I've thought about it some more and have run some more tests, I suspect I might have misunderstood how everything works. What farinha says might very well be the case -- that I've confused bits with bytes. The Speakeasy test I ran tonight showed a download speed of 1441 Kbps, which then translates to 180.1 KB/sec. And the consistency in speeds (e.g. nothing ever more than roughly 180 KB/sec) across two service providers would also suggest that the KB/sec rate I'm getting is pretty good for the overall speed of the connection I'm paying for. I think I really was mixing up bytes with bits.

Also, I found a few sites that do conversion of numbers from MB/sec to Mbps, Kbps, etc., and when I run the numbers, it all makes more sense.

So, creating this thread was a real lesson. I'm glad I started it because I might have kept assuming that my download speeds were woefully inadequate.

By the way, I dumped Verizon DSL for Time-Warner/Road Runner because I did have a real problem with my connections with Verizon -- for months, it would consistently go down between approximately 9-10 AM in the morning every day and then remain down (or intermittent) for a few hours after that. I had multiple technicians out to the house, I spent countless hours on the phone with tech support, and Verizon rewired the lines coming into the house -- all to no avail. They never could locate the source of the outages (or they weren't really trying hard enough, except for one technician who made it his mission to try to figure it out). I don't expect Road Runner to be perfect, but so far, so good.

 

[moshkid213]

Most ISP's advertise in megabits per/second

Reason why?

8Megabits per second looks faster and more appealing then 1 Megabyte per second. There is 8 bits in a byte, so devide whatever speed they give you by 8 and thats your actual bandwidth cap. (Hence the word cap, doesn't mean you will download that speed)

Kinda falls under the grounds of trying to have a gigabit network, EVERY piece of layer 1 equipment needs to be gigabit certified, Example, Cat 6 all the way through, (Speed tests verify that Cat5E can reach gigabit speeds but is not certified to do so) Then all your switches need to be gigabit as well, Then, your routers as well! But guess what, a big business is going to have subinterfaces on their router to run VLANS so they don't have as many collisions, which (Correct me if I'm wrong) will split bandwidth on that interface out to the different networks.

Networking is kinda, hook it up, configure it, hope for the best. In a large network its hard to tell where your bandwidth cap is, traceroutes can show you where your packets are being delayed the most, thats about it, but that can lead to just a router running hot, Etc.

Hope this helped. Sorry If I went overboard.

 

[drmike]

moshkid213, thanks for that explanation -- not overboard at all. I always wondered why the estimated speeds were always bits, not bytes. And I agree; even in my limited experience with networks, they all do seem to involve hooking things up and then just hoping for the best.