Re: detecting connection speed

On Wed, Dec 4, 2013 at 2:51 AM, James Graham <james@hoppipolla.co.uk> wrote:

> On 04/12/13 05:05, Puneet Mohan Sangal wrote:
>
>> James, I understand it's a difficult problem to solve, however it will
>> be a very useful dimension when conducting analysis on performance data.
>> May be we should start with what the possible solutions could be, and
>> then consider the pros & cons?
>>
>
> You should always start with use cases, because they form the basis for
> determining what the pros and cons of possible solutions are. If possible
> you should also back up use cases with evidence that they are real problems
> e.g. people using workarounds to try and solve the use case in the absence
> of a built in solution. In this case that might be e.g. authors timing
> resource downloads to try to estimate connection speed.
>

Timing resource downloads is a more direct measure of the only (or at
least main) property of interest - the achievable performance on the
current network connection, from a particular server and at the current
time. Adaptive streaming applications, for example, make extensive use of
resource download timing - directly or indirectly - to determine the
appropriate video rate to request.

Historical performance information can be useful to bootstrap this process,
so that clients can determine an appropriate video rate from the start, for
example.

So, I'm not sure what "connection speed" is or why it is interesting ? I
can see that knowing about the network technologies involved in the last
mile can be useful for post-hoc analysis of the performance and properties
of those technologies, but this is more than could be provided by a Network
API at the client: one needs to know about ISPs and the technology they
have deployed, the end user's tier of service with their ISP, the home
networking setup etc.

...Mark

Received on Wednesday, 4 December 2013 16:30:10 UTC