Running an internet speed test on my mobile phone indicates that I get a much faster download speed than my ISP broadband connection (yes, that's how bad my "broadband" is). However the ping is far longer/slower on my mobile connection.

I would've thought ping and speed would go hand in hand.

I understand the concepts of ping, i.e. a measurement of a round trip between a source and destination.

So would this suggest that the mobile connection is "further" from its destination, but "wider" and capable of higher throughput?


The easiest way to imagine this is to think of a river - the water as data and a floating rubber duck as your ping (albeit only in one direction).

A river has two relevant properties - size (width & depth) and speed (flow rate). A large river may move a lot of water, but it doesn't necessarily mean the flow rate is high (so your duck might take a long time to get from one point to another). Likewise a small river may not move much water, but it can flow rapidly.

Of course, you can get small rivers which flow slowly, and large rivers which flow rapidly, and it is the latter which you ideally want when it comes to broadband. The two properties are however not necessarily directly linked.

A further issue is that routers on each hop of the way to and from your destination can delay the data transfer. To continue with the analogy, imagine that the river has a number of locks along its route. These locks can hold up your duck until it is ready to let your duck proceed to the next part of the river.

Disclaimer: putting your modem into a big fast river won't help.

  • 1
    +1, as mentioned here latency and bandwidth are not the same thing and are not necessarily linked. For example satellite links are very high latency (high ping) yet can also be very high bandwidth while a direct LAN connection will be very low latency with high bandwidth. Direct wire connections will always be lower latency than radio connections due to conversions required. Radio connections tend to compensate for slow links by transferring larger amounts of data in one go. – Mokubai Dec 28 '12 at 0:18
  • 2
    This river water analogy is dangerous, as it perpetuates the OP's misconception that the mobile connection is ""wider" and capable of higher throughput" but somehow slow. The problem with most analogies is that digital communications is often packetized, and can be buffered (or delayed) during transmission stages. So there's a stop'n'go characteristic that is rarely mentioned. – sawdust Dec 28 '12 at 0:57
  • 1
    @sawdust Hopefully locking up some ducks helps? – Graham Wager Dec 28 '12 at 1:09
  • Thanks Graham +1 from me. Good point on the packetized nature. – Alex KeySmith Dec 28 '12 at 18:19

I think you have the concepts correct. I have the same results in terms of a high bandwidth but with high ping/latency... It has to do with the way in which the data is transferred... Cell connections are much less efficient than that of a true broadband connection.


I would've thought ping and speed would go hand in hand.

No, that's why most speed tests report these separately.
The ping requires a response from the server before the second half of the "round trip" can continue. So the ping time is the sum of the upstream transmit time, time for the server to respond, and the downstream transmit time.

So would this suggest that the mobile connection is "further" from its destination, but "wider" and capable of higher throughput?

These signals travel at the speed of light, so the distance between source and destination would have to be very large before it's a factor. Your mobile phone is probably serviced by a tower within a few miles. Even a (low orbit) sat phone transmission is less than a thousand miles.

Almost all long-distance transmission schemes, including the two methods you are comparing, use serial transmission. They might employ a modulation schemes (digital information over analog channel) that transmits several bits per symbol (e.g. 3 bits for 8VSB or 8 bits for QAM256). But this is accounted for in the bits per second specification for the transmission rate.

You simply cannot infer request&response throughput performance based on unidirectional transfer rate. You need to account for server response time, the number of intermediate relays or hops that may occur (store & forward retransmissions that are fully buffered or cut through that add only a small delay) and protocol overhead.

  • 1
    "These signals travel at the speed of light, so the distance between source and destination is not a factor." This is only truth for a small part of the path (between the mobile and the BTS). The distance that should be accounted is the topological distance in terms of hops, and that is a factor. – fmanco Dec 28 '12 at 0:26
  • @criziot - Yes, the number of hops is significant. I made an edit to include that concept just as you completed your comment. – sawdust Dec 28 '12 at 0:29
  • Thanks for the extra technical details +1 from me. But don't worry, when I sat further I didn't mean physical distance. – Alex KeySmith Dec 28 '12 at 18:17

Major phone networks frequently use QoS to deprioritize ICMP and other traffic irrelevant to service performance.

  • That's very interesting Josh. – Alex KeySmith Jul 31 '13 at 9:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.