[sldev] Cache politics: performance vs obfuscation

Tateru Nino tateru.nino at gmail.com
Tue Jun 10 23:47:03 PDT 2008



Loom wrote:
>
> On 11/06/2008, *Matthew Underwood* <sakkaku at gmail.com 
> <mailto:sakkaku at gmail.com>> wrote:
>
>     On Tue, Jun 10, 2008 at 7:55 PM, Loom <loom at loomiverse.net
>     <mailto:loom at loomiverse.net>> wrote:
>     >
>     >
>     > On 11/06/2008, Ann Otoole <missannotoole at yahoo.com
>     <mailto:missannotoole at yahoo.com>> wrote:
>     >>
>     >> Let us ensure we have the priorities straight.
>     >> With Time Warner and other cable modem service providers
>     preparing to
>     >> implement bytes transferred caps it is imperative that the
>     cache operation
>     >> overall be improved to ensure there is no repetitive
>     downloading of bytes.
>     >>
>     >> It simply will not matter if the bytes are obfuscated if the
>     population of
>     >> Secondlife is gutted by a large scale *perceived inability* to
>     afford to
>     >> participate.
>     >>
>     >> Save SL first. Then worry about the rest.
>     >> If SL dies because of consumers *not being aware of their
>     choices in
>     >> service providers* then nothing really matters does it?
>     >
>     > I realise that you have used the word "perceived" in there but I
>     would like
>     > to suggest that usage caps are not going to kill secondlife.
>     >
>     > Personally, I don't agree with the idea of usage caps, however
>     in Australia
>     > we have lived with them for quite a while.
>     >
>     > From the march 2008 economic stats, there were 12,245 active
>     Australian
>     > users or about 0.057% of the population, who clocked up an
>     average of 47
>     > hours each - using usage capped internet connections.
>     >
>     > Compare that with the US where 194,899 active users (0.064% of the
>     > population) clocked up an average of 59 hours each.
>     >
>     > At worst, SL may take a hit in usage, but suggesting that it
>     will die is
>     > imho an extreme view which is more in the line of fearmongering than
>     > rational analysis.
>     >
>     > Loom
>
>
>     The idea is to plan ahead to avoid running full on into a wall.  There
>     is no point in bickering over weather or not people are able to play
>     SL, it is to realize that people ARE capped and MAY be impaired by SL.
>       Why not improve the caching such that the asset servers are hammered
>     less, people have to use less bandwidth and therefor have a more
>     enjoyable experience.
>
> No argument there, I'm simply point out that an "OMG usage caps will 
> kill SL" attitude is not necessarliy a reasonable stance to take.  
>
> A simple and I believe strong argument for both better cache and 
> texture decoding code is that it has to be faster than downloading an 
> asset from somewhere else on the internet every single time it is 
> needed.  The fatser the code, the better the SL experience for anyone. 
>  In a sense, the paranoia about caps being introduced in the US will 
> potentially make SL better for everyone if it means that cache 
> performance is improved.
>

Caps would certainly cause each user to reevaluate how they spend their 
network traffic, in what proportion and how often.


More information about the SLDev mailing list