Comparing Transmission Speeds

Comparing Transmission Speeds

  • Comments 9

Have you ever wondered how various transmission speeds relate to each other? Or have you ever stopped to consider how USB 2.0 compares to Fast Ethernet? I frequently encounter people—even technical people—who don't seem to get it. I was listening to a podcast (to remain unnamed) just yesterday where the participants were reviewing a portable audio/video device. After commenting that the device supported both USB 2.0 and Firewire connections and noting the relatively large size of video files, one of the speakers ranted: "but I wish it had an Ethernet port." To which I mentally responded: "why...so you can transmit your data at one fifth the speed!?!?" The clear implication was that an Ethernet connection would provide the faster solution.

I also get grief from people because I almost always plug a network cable into the back of my laptop, even if I'm in wireless range. I'll grant that a wireless connection is the ultimate in convenience, and it's often good enough to be faster than your broadband connection at home, but if you're transferring files of any size, let me be the first to tell you that wireless will almost always lose against a standard Fast Ethernet connection. It's important to point out that there's a big difference between theoretical and actual transmission speeds. In the case of a wireless connection, achieving 50% of the theoretical maximum is quite common, especially in a congested area. That means that if you're running 802.11b, you may be limiting yourself to 5Mbps (of course, your mileage may vary). So, if you have an 8Mbps broadband connection to the internet, your wireless connection becomes the bottleneck.

If you don't believe me, I encourage you to copy a large set of files over your wireless connection, then try the same experiment over a wired connection. You won't need the precision of a stopwatch for your test. Chances are extremely good that you'll see a very noticeable speed difference without doing any measurements. I like to think of it this way: wireless = convenience; wired = speed. By the way, if you are in a congested area and you're a relatively technical person, I'd recommend re-running this same test for each of the available transmission channels on your wireless access point. You should be able to find a channel that provides you with less interference and that results in faster transfers.

I've put together a quick chart that compares the theoretical maximum transmission speeds of some common technologies. All of the speeds are measured in Mbps, which is megabits per second. I only mention this because I've seen many people confuse Mbps and MBps. Lower-case "b" means "bits," and upper-case "B" means "bytes." Click on the chart for a larger version.

Leave a Comment
  • Please add 3 and 3 and type the answer here:
  • Post
  • I am missing the SATA/PATA and fiberchannel interfaces from this comparation :(
    But anyway, it's nice :)
  • I liked the chart quite a bit. I hadn't realized how slow SCSI-1 was in relative terms. It'd be interesting to see the same data for ATA and SATA interfaces too.
  • A question worth millon dollars:
    Why then you plug network cable you have to use that connection instead of wireless. Why this is impossible to combine them ?

  • Great chart, Mike.  I'm going to forward a link to somebody who recently teased me about using a cable in a meeting.

    I agree, wireless is a bad choice if you're moving a lot of data.  I've seen copy operations over wireless where it would have been faster to drive across town to a wired connection, copy the file there, and drive back.
  • Transmission speeds are only part of the story. Protocol is a key part; for example notice how USB2 (Hi Speed) is faster than Firewire by 80 Mbps. If you do the file copy test you will find that Frewire will copy faster that USB2. One reason is that Firewire protocol uses a look ahead algorithm where USB2 doesnt.
  • Great point, Mojo. That's why I was careful to note that these are *theoretical* maximum speeds and that your mileage may vary. But, the chart is useful as an overall way to understand the relative differences.

    The other piece of the puzzle is the speed of the device. Just because I have a fat pipe between the device and the computer doesn't mean that the device can deliver/receive data at full speed.

    As always, it's best to test, test, test. :)
  • Agreed - I move GBs of data on daily  basis  in my job and always rely on my 100mbps wired connection to do the job; wireless is great for the mundane email, etc., but can't beat the 15-20x bandwidth of the wired connection.

    The fairly widespread ignorance on this topic confounds me.

    On a slightly different note, I can't wait until Intel and others start baking the TCP/IP stack on-chip so we can then take the load off of Windows.
  • "Why this is impossible to combine them?"

    The problem is with IP (as in, TCP/IP). Each of those devices has a different IP address, so you can't use them both for a single connection.

    PPP (a layer below TCP/IP) supports bundling multiple physical connections together (called MultiLink). This was originally added for things like ISDN, where you had two 64k data connections which you combined at the lower layer into a single 128k data connection (TCP/IP always only ever saw it as a single logical connection). You can even make or break pieces of the connection, so you could use your ISDN to accept phone calls, halving your bandwidth while on the phone, and then re-establishing full speed when you're done.

    In theory, if you had an endpoint and the right mix of software, you could use PPPoE along with MultiLink to physically combine your connections. It would provide some nice features, too, like transparently transitioning you from wired to wireless when your wire goes away, without having to dump all the connections and restart them. In reality, the benefit is small enough that nobody has bothered with it. The inconveniences of switching media are relatively small.

Page 1 of 1 (9 items)