Re: Running _every_ page through WebCat ?
This WebDNA talk-list message is from 1997
It keeps the original formatting.
numero = 10717
interpreted = N
texte = >There have been some messages from folks who serve all their pages >through WebCat. This certainly isn't what we had originally intended to >do, just serve up catalog and ordering type stuff, but the sheer power of >WebCat is making it a rather interesting proposition. > >Before we investigate that possibility in more detail, does anyone have a >quantitative estimate of how it would impact overall performance. Even a >qualitative estimate would be helpful. We serve up less than 10,000 files >per day (with occasional exceptions :-), so we're not a high volume site. >However, performance is still an issue as it relates to what the >individual visitor sees. > >Any thoughts ? > >Thanks, > > >-Sven The below is more than you asked for... but others might be interested in the same data.Used to use netcloak and rushhour on our site. What we did is a crude method of monitoring speed. Using Netscape 3.0, we would clear the cache on the browser before the load of the page. Using a stop watch we would record the time to load. And then repeat the process 10 times on 6 different pages on 4 different nights. All at about the same time (8 pm) We also pinged out sight to check for packet loss and turn around time to make sure the connection via our dial up wasn't an issue. We used a netcom.com dial up account that was not in our ip range, vs using one from our isp which is. IE netcom is 199.xxx.xxx.xxx and my isp is 207.xxx.xxx.xxx.I don't have the data with me, since the tests were all done from my home office on an 8500/180 and a supra 28.8. The server CPU is 7100/66 with 72 megs of ram, Mac/TCP not OT. Memory control panel disc cache set wide open at 7680, and virtual turned off. Webstar file info cache is set at 1024 and buffer size is set at 1500. DNS is done on a 6100/66 right next to the 7100. Connection is ISDN (128) thru an Ascend P50 router. Dropped rush hour first and went to Data Cache plug-in provided by webstar and gave it 6 megs and then set all .jpg and .gif to be served thru the data cache plugin. This seemed to make our pages serve slightly faster, but also smoother. Rush hour served images seemed to hang once and awhile. Sort of like a pause, then then go again. With the test page that comes to mind, the speed increase was something like 15 to 20 percent average over the 10 loads over the 4 nights. Other stuff we run is calset.acgi, netfax.acgi forms.acgi and netfinder.acgiThen we took the same 10 pages and copied them and change the extension to .tmpl. So now I had 2 identical documents on the server. One .html and the other .tmpl. The .html's in the test were set in netlcloak cache to load up durning the start up process. We conducted the same tests as above. Off the top of my head I think the numbers were like around 10 to 15 percent faster thru webcatalog. I know they were enough faster that two of us, spent the better part of last Saturday, removing all reference to netcloak and rush hour on over 500 pages. Now I could pull together the real data from home and post it to you if needed Monday. BTW we has webstar getting 16 megs and if I do a [freememory] it shows webcatalog at around 6 megs to start with. And 6017808 right at this instant. All of our site now goes thru WebCatalog. What I find weird about this whole site change over is the e-mail comments we get asking what we did from visitors to speed the thing up. Never got any that said it was slow. But for somebody to click on send mail to webmaster and comment about it being faster surprised me.It is my humble opinion the Webcatalog2.0b12 and up are far more stable the the latest netcloak beta's. Webcatalog has never crashed my site, unless it was somebodies syntax error. And with some creative thinking you can do just about every thing netcloak does in webcatalog. And it many cases more. We now sort our page counters by date or area. We are not a shopping site yet (soon) we are more of a data base system. Create some templates and let others do the input and forget the html coding for the routine pages that get updated sometimes 3 or 4 times a week.The one command that we over looked when we did these tests is [elaspedtime]. Put one at the top and one at the bottom of a page. We are very sensative to undo eye candy on any of our pages. We figure the faster our pages load the longer someone is going to prowl our site. I know that if some site loads very slow I hate going there and avoid it as much as possible when using dial up.I am curious if Grant or anybody else has used pounder or a like program and has real data, not the seat of the pants tests we did?PS: one thing we did recently was replace our router from Netopia to Ascend P-50. That made our site sing both on the Lan and the Wan. I have 26 users hanging on this thing and a AIX RISC 6000 that is the work horse of our business applications and file sharing. And a 6150/66 that serves out pages for our Sales reps, Plus more printers and plotters then I care to count :).More than you asked for, but others might like to know what we did and or I would apreciate comments about the above from others.=============================================== Gary Richter PanaVise Products, Inc. 7540 Colbert Dr. Reno, Nevada 89511 Ph: 702.850.2900 Fx: 702.850.2929 Email: grichter@panavise.com http://www.panavise.com ===============================================
Associated Messages, from the most recent to the oldest:
>There have been some messages from folks who serve all their pages >through WebCat. This certainly isn't what we had originally intended to >do, just serve up catalog and ordering type stuff, but the sheer power of >WebCat is making it a rather interesting proposition. > >Before we investigate that possibility in more detail, does anyone have a >quantitative estimate of how it would impact overall performance. Even a >qualitative estimate would be helpful. We serve up less than 10,000 files >per day (with occasional exceptions :-), so we're not a high volume site. >However, performance is still an issue as it relates to what the >individual visitor sees. > >Any thoughts ? > >Thanks, > > >-Sven The below is more than you asked for... but others might be interested in the same data.Used to use netcloak and rushhour on our site. What we did is a crude method of monitoring speed. Using Netscape 3.0, we would clear the cache on the browser before the load of the page. Using a stop watch we would record the time to load. And then repeat the process 10 times on 6 different pages on 4 different nights. All at about the same time (8 pm) We also pinged out sight to check for packet loss and turn around time to make sure the connection via our dial up wasn't an issue. We used a netcom.com dial up account that was not in our ip range, vs using one from our isp which is. IE netcom is 199.xxx.xxx.xxx and my isp is 207.xxx.xxx.xxx.I don't have the data with me, since the tests were all done from my home office on an 8500/180 and a supra 28.8. The server CPU is 7100/66 with 72 megs of ram, Mac/TCP not OT. Memory control panel disc cache set wide open at 7680, and virtual turned off. Webstar file info cache is set at 1024 and buffer size is set at 1500. DNS is done on a 6100/66 right next to the 7100. Connection is ISDN (128) thru an Ascend P50 router. Dropped rush hour first and went to Data Cache plug-in provided by webstar and gave it 6 megs and then set all .jpg and .gif to be served thru the data cache plugin. This seemed to make our pages serve slightly faster, but also smoother. Rush hour served images seemed to hang once and awhile. Sort of like a pause, then then go again. With the test page that comes to mind, the speed increase was something like 15 to 20 percent average over the 10 loads over the 4 nights. Other stuff we run is calset.acgi, netfax.acgi forms.acgi and netfinder.acgiThen we took the same 10 pages and copied them and change the extension to .tmpl. So now I had 2 identical documents on the server. One .html and the other .tmpl. The .html's in the test were set in netlcloak cache to load up durning the start up process. We conducted the same tests as above. Off the top of my head I think the numbers were like around 10 to 15 percent faster thru webcatalog. I know they were enough faster that two of us, spent the better part of last Saturday, removing all reference to netcloak and rush hour on over 500 pages. Now I could pull together the real data from home and post it to you if needed Monday. BTW we has webstar getting 16 megs and if I do a
[freememory] it shows webcatalog at around 6 megs to start with. And 6017808 right at this instant. All of our site now goes thru WebCatalog. What I find weird about this whole site change over is the e-mail comments we get asking what we did from visitors to speed the thing up. Never got any that said it was slow. But for somebody to click on send mail to webmaster and comment about it being faster surprised me.It is my humble opinion the Webcatalog2.0b12 and up are far more stable the the latest netcloak beta's. Webcatalog has never crashed my site, unless it was somebodies syntax error. And with some creative thinking you can do just about every thing netcloak does in webcatalog. And it many cases more. We now sort our page counters by date or area. We are not a shopping site yet (soon) we are more of a data base system. Create some templates and let others do the input and forget the html coding for the routine pages that get updated sometimes 3 or 4 times a week.The one command that we over looked when we did these tests is [elaspedtime]. Put one at the top and one at the bottom of a page. We are very sensative to undo eye candy on any of our pages. We figure the faster our pages load the longer someone is going to prowl our site. I know that if some site loads very slow I hate going there and avoid it as much as possible when using dial up.I am curious if Grant or anybody else has used pounder or a like program and has real data, not the seat of the pants tests we did?PS: one thing we did recently was replace our router from Netopia to Ascend P-50. That made our site sing both on the Lan and the Wan. I have 26 users hanging on this thing and a AIX RISC 6000 that is the work horse of our business applications and file sharing. And a 6150/66 that serves out pages for our Sales reps, Plus more printers and plotters then I care to count :).More than you asked for, but others might like to know what we did and or I would apreciate comments about the above from others.=============================================== Gary Richter PanaVise Products, Inc. 7540 Colbert Dr. Reno, Nevada 89511 Ph: 702.850.2900 Fx: 702.850.2929 Email: grichter@panavise.com http://www.panavise.com ===============================================
grichter@panavise.com (Gary Richter)
DOWNLOAD WEBDNA NOW!
Top Articles:
Talk List
The WebDNA community talk-list is the best place to get some help: several hundred extremely proficient programmers with an excellent knowledge of WebDNA and an excellent spirit will deliver all the tips and tricks you can imagine...
Related Readings:
two of the same named db's open at once (1998)
RAM variables (1997)
PCS Frames (1997)
[OT] Bookmarking code (2003)
PCS Emailer's role ? (1997)
Multi or auto submit (1999)
SMSI (Scott) clues?? Bug fixes for next WebDNA release (2005)
WebCatalog for guestbook ? (1997)
RAM variables (1997)
Showif, Hideif reverse logic ? (1997)
Help! WebCat2 bug (1997)
uninstalling webcatalog (2000)
Appendfile memory usage (redux) (2003)
FM PRO Compatibility Issue - Single Database w/o Conversions (1997)
WebMechant (2000)
[shownext] and sort (1998)
Emailer (1997)
problems with 2 tags (1997)
Summing fields (1997)
PCS Frames (1997)