Web site speed - the only way to REALLY know...

["David Anumudu" <David.Anumudu@xxxxxxx>]

Hi Mark

I am a lurker on Mark's HA newsgroup and just picked up the thread about
web
site performance testing.

Important stuff first - I have just given the builders my plans for
installing cat5 for the new house I am moving in to (though I have to admit
that I am not running much cat5, I see it only as a means for distributing
video around the house using kat5, phone and data I will do wireless, and
for audio I am driving speakers driven from amps in node 0). I will also be
fitting Comfort, and LD controlled lighting... Roll on December which is
when I move in..

Anyway, back to the question about web site performance. I work for a
company that sells load and scalability testing software, which basically
allows you to measure the real performance of your application by recording
the users activities in a browser, which are then saved in to a fully
editable JavaScript file, which is then fed in to a load generator or
multiple load generators (PCs running NT or Linux). Each PC can simulate
multiple virtual users, allowing you to therefore simulate tens, hundreds,
thousands or even tens of thousands of concurrent users accessing your
application, while giving you real time stats of both the performance that
the simulated users are seeing, and also the workload on various parts of
the application architecture.

The recommendations that people have made about optimising graphics are
valid, but they are only one part of the story, particularly when the
application has lots of dynamic content served up by back end databases,
application servers etc.

In summary, the ONLY way to know how your web application will perform is
to
measure it under load, I have lost count of the number of customers I have
been to (international corporates who have spent in some cases millions of
dollars developing and deploying web applications) who had assumed that
their application would easily handle X hundred users, and when actually
tested they have collapsed less than a dozen users!

Without it sounding like a shameless plug - if you are interested then let
me know and I can get a 'no commitment on site demo' arrangement where you
can see it running for real.

regards

David Anumudu

----- Original Message -----
From: "Mark Harrison" <Mark.Harrison@xxxxxxx>
To: <ukha_d@xxxxxxx>
Sent: Friday, October 12, 2001 11:36 AM
Subject: RE: [ukha_d] OT: Web site speed


> Thanks again.
>
>
> Mark Harrison
> Head of Systems, eKingfisher
>
>
>
> -----Original Message-----
> From: Nick Shore [mailto:nick.shore@xxxxxxx]
> Sent: 12 October 2001 11:27
> To: ukha_d@xxxxxxx
> Subject: RE: [ukha_d] OT: Web site speed
>
>
> Its worth cross checking that when image tags are used that the width
> and height are set correctly.
> Also worth playing with quality of jpegs and trying pngs.
>
> Nick.
>
>
> -----Original Message-----
> From: Mark Harrison [mailto:Mark.Harrison@xxxxxxx]
> Sent: Friday, October 12, 2001 11:15
> To: ukha_d@xxxxxxx
> Subject: RE: [ukha_d] OT: Web site speed
>
>
> Graham
>
> Thanks for this. The brand-marketing is the utter lead here, hence the
> insistence on exactly the right "corporate" fonts which, as
you say,
> effectively means graphics.
>
> As you suspected, there's a _lot_ of back-end integration on this site
-
> I don't see how we could run a site with 80,000 different items on it
> without ;-)
>
> Your point about "reloading fast on return" is excellent!!!
>
> Mark Harrison
> Head of Systems, eKingfisher
> -----Original Message-----
> From: Graham Howe [mailto:graham@xxxxxxx]
> Sent: 12 October 2001 11:08
> To: ukha_d@xxxxxxx
> Subject: RE: [ukha_d] OT: Web site speed
>
>
> If you are talking about www.diy.com then I have to say I thought it
was
> pretty quick for a 'marketing driven'  site. I cleared out my cache
and
> loaded the page in under 20 seconds on ISDN 64k line. A lot of the
> content that usually slows sites down (like flash, sound, animated
gifs
> etc etc) are not present which is good, so the only real performance
hit
> is the graphics. Most of the graphics seem to be small gifs with only
a
> few jpgs so that is not bad.
>
> I don't believe that the marketers would allow you to strip off the
> photos, but that would certainly speed up the page a bit. There is a
> fair amount of use of gifs to represent words and of course if those
> words were instead created using style based text then it would be
> faster. But the marketers are again likely to put their foot down
> because they will want the specific fonts and those might not be
> possible to accurately reproduce on the page. The other problem will
be
> that even the same font can be displayed at slightly different sizes
by
> different browsers, so to force consistent appearance across browsers
it
> is often necessary to resort to graphics. I would say that they should
> check out the competition and see how their pages load, for example
> www.homebase.co.uk is much slower as they have several photos on the
> front page, likewise the wickes site is pretty slow.
>
> In summary I'm nt so sure there is anything that can be done without
> removing a fair bit of the look of the page, but you may want to see
if
> the marketers would be willing to lose the fancy fonts and have all
> words and numbers displayed as text (with appropriate styling) rather
> than graphics.
>
> Regards
>
> Graham
>
> P.S. if there is any code or database interaction involved in actually
> building the page on the server (looking at the source I suspect there
> is), then you might want to look for any inefficiencies there as well.
> But to be honest it loads almost immediately on a return visit, so it
> does appear that all delay is down to graphics.
>
> > -----Original Message-----
> > From: Mark Harrison [mailto:Mark.Harrison@xxxxxxx]
> > Sent: 12 October 2001 10:18
> > To: ukha_d@xxxxxxx
> > Subject: [ukha_d] OT: Web site speed
> >
> >
> > Hi all,
> >
> > An off-topic request to all you Web gurus.
> >
> > B&Q are concerned that there web site's home-page takes too
> > long to load
> > (it does!)
> >
> > So the question is, what would be your recommendations for making
it
> > load faster? Bearing in mind that marketers control the look ;-)
> >
> > Not sure if this should be on-group, so feel free to email me
> > direct if
> > you prefer!
> >
> > Mark Harrison
> > Head of Systems, eKingfisher
> >
> > ------------------------ Yahoo! Groups Sponsor
> > ---------------------~-->
> > FREE COLLEGE MONEY
> > CLICK HERE to search
> > 600,000 scholarships!
> > http://us.click.yahoo.com/Pv4pGD/4m7CAA/ySSFAA/IBOolB/TM
> <http://us.click.yahoo.com/Pv4pGD/4m7CAA/ySSFAA/IBOolB/TM>
> > --------------------------------------------------------------
> > -------~->
> >
> > For more information: http://www.automatedhome.co.uk
> <http://www.automatedhome.co.uk>
> > Post message: ukha_d@xxxxxxx
> > Subscribe:  ukha_d-subscribe@xxxxxxx
> > Unsubscribe:  ukha_d-unsubscribe@xxxxxxx
> > List owner:  ukha_d-owner@xxxxxxx
> >
> > Your use of Yahoo! Groups is subject to
> > http://docs.yahoo.com/info/terms/
<http://docs.yahoo.com/info/terms/>
>
> >
> >
> >
> >
>
>
> For more information: http://www.automatedhome.co.uk
> <http://www.automatedhome.co.uk>
> Post message: ukha_d@xxxxxxx
> Subscribe:  ukha_d-subscribe@xxxxxxx
> Unsubscribe:  ukha_d-unsubscribe@xxxxxxx
> List owner:  ukha_d-owner@xxxxxxx
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of
> <http://docs.yahoo.com/info/terms/>
Service.
>
>
>
> Yahoo! Groups Sponsor
>
> ADVERTISEMENT
>
>
>
> <http://rd.yahoo.com/M=194081.1637497.3177299.1261774/D=egroupweb/S=1705
> 041992:HM/A=793313/R=1/*http://www.ediets.com/start.cfm?code=3258>
>
> <http://us.adserver.yahoo.com/l?M=194081.1637497.3177299.1261774/D=egrou
> pmail/S=1705041992:HM/A=793313/rand=720109004>
>
> For more information: http://www.automatedhome.co.uk
> <http://www.automatedhome.co.uk>
> Post message: ukha_d@xxxxxxx
> Subscribe:  ukha_d-subscribe@xxxxxxx
> Unsubscribe:  ukha_d-unsubscribe@xxxxxxx
> List owner:  ukha_d-owner@xxxxxxx
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of
> <http://docs.yahoo.com/info/terms/>
Service.
>
> _____________________________________________________________________
> This message has been checked for all known viruses by the
> MessageLabs Virus Scanning Service. For further information visit
> http://www.messagelabs.com/stats.asp
>
>
>
>
>
> For more information: http://www.automatedhome.co.uk
> Post message: ukha_d@xxxxxxx
> Subscribe:  ukha_d-subscribe@xxxxxxx
> Unsubscribe:  ukha_d-unsubscribe@xxxxxxx
> List owner:  ukha_d-owner@xxxxxxx
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>