Re: [Myexperiment-discuss] slowExperiment

[Wim De Smet]

Hi,

Paul Fisher wrote:
> Hi,
>
> I have noticed recently that the time taken to log into myExperiment 
> and searching/using it has significantly increased. I'm not sure if 
> this is just my machine or not.
> Is there any problems with the server at all, that could be causing this?
>
> I've already mentioned this to Jits, but I'll post it here too as a 
> record.
> A friend of mine told me a while ago about how Google speed up the 
> return of search results by removing all un-necessary white space from 
> the returned page, since whitespace still has to be encoded and 
> returned. As an example:
> [...]
>     
> Looking at myexperiment, however, for a search on "microarray" gives 
> me a nicely formatted page, but with a hurrendous amount of whitespace 
> in there.
> Could this be a subtle cause of the lag?
>
> Please do correct me if I'm wrong here. I may have confused myself, as 
> I often do.
> Nurse.... nurse....!!!
Not a myexperiment developer (well, not really) but I believe you're 
only partly right. 99% of the time any slowness you're experiencing is a 
problem on the backend (i.e. server code not executing fast enough, 
database queries taking some time, that sort of stuff). Whitespace can 
sometimes trigger some really weird browser bugs (in, what else, IE6 
mainly) but shouldn't really pose a noticeable difference, especially 
since it's usually being compressed. Spaces do add up, but generally not 
enough that you'd notice.

The real benefit to google in reducing whitespace is that they reduce 
bandwidth considerably. When you start serving pages at the volume they 
are, that extra whitespace starts adding up in your bandwidth bills. 
Even shaving off a couple of bytes of every page delivered can save them 
thousands to millions of dollars, depending. Myexperiment has a lot of 
users but I don't think they're seeing that sort of problem just yet.
> regards,
> Paul.
> p.s. would this also reduce the myExperiment "carbon foot-print" by 
> limiting the amount of data sent back to the user?
Well, more data means processor need to execute a couple more cycles, 
which might add up to a couple of Kwh per year, maybe. (I'd have to do 
some back of the envelope calculations, but that sounds really 
conservatively high to me) So yes, I guess so.