For the record, I agree with rsohan.
Picking up from Mr Okoth,
Assuming I'm transferring 10 Mbps.
On a 1Mbps link, over one second, I can only put 1 Mb of data on the wire, however, for the 10Mb link, I can put 10 Mb of data on the wire.
Hence the 10Mb link will be approximately 10x faster.
Of course in reality though, other factors such as protocol choice eg TCP vs UDP, window size, packet size, propagation delay, cable length, transimssion media etc will have their effect.
A good way to look at how latency affects throughput, is to think from the TCP point of view; The device has to wait for an acknowledgement before putting more data on the wire. = wasted time if the acknowledgement is not so important.= Lower throughput.
If my packet is too large, the first hop has to receive the full packet before it can start transmitting it to the second hop.= wasted time =lower throughput.
If I halve the packet size, the first hop can receive the first packet and start transimtting it before receiving the second packet.= cumulatively saved time = higher throughput.
But only to a certain point before the overheads become a limiting factor. ( Think ATM cell size).
It really is a complicated but interesting topic. IMHO :-)
On Mon, May 7, 2012 at 8:36 AM, Gregory Okoth
<gregory.okoth@gmail.com> wrote:
@Sohan,
In your earlier example, some factors were varying. I would like you to give an explanation, keeping all factors constant i.e
- Lenght of cable is same,
- Identical physical x-tics of the tramission media, and
- a similar data size being transmitted across
expect of course for the link capacities which will be a comparison for 1Mbps and a 10Mbps link.
Would the same explanation suffice?
On Sun, May 6, 2012 at 7:47 AM, Stephen Munguti
<kamitu.sm@gmail.com> wrote:
Throughput of a network (Which i presume is what you are referring to as speed) depends on
Your presumption is erroneous. When I mentioned speed I was explicitly clear to mention it is the speed of propagation of light and not the maximum attainable bandwidth of the network.
latency, the application (e.g TCP/IP window sizes) and bandwidth (Maximum packets one can send across a link under ideal conditions)
Which is why you will notice I go to great pains to highlight the fact that I assume otherwise "identical" networks. This serves to encompass and abstract away everything else apart from the parameters in question (link BW and length)
The concepts form a lengthy explanation, but you cannot separate throughput and latency
Which was my exact point two emails back. So either the point of this email was to agree with me or I have misunderstood. Given your initial sentences I don't think it was the former. Please elucidate further.
With kind regards
--
Regards,
Greg
--------
Life is not a rehearsal, you only live once!
_______________________________________________
Skunkworks mailing list
Skunkworks@lists.my.co.ke
------------
List info, subscribe/unsubscribe
http://lists.my.co.ke/cgi-bin/mailman/listinfo/skunkworks
------------
Skunkworks Rules
http://my.co.ke/phpbb/viewtopic.php?f=24&t=94
------------
Other services @ http://my.co.ke