Re: WebTEN vs webSTARtopic
This WebDNA talk-list message is from 1998
It keeps the original formatting.
numero = 18917
interpreted = N
texte = >Any comments about the recent press stating that WebTEN trounces WebSTAR>even at low volumes?>>I am using WebSTAR 3.0.... Should I sidegrade to WebTEN?>>http://www4.zdnet.com/pcmag/features/webserver98/bench.html>>http://macweek.zdnet.com/ns-search/1223/rv_web.html?NS-search-set=/35ad1/aaa>a000oqad1f5a&NS-docset=0&>Ahh but wait.. this is from todays firestite digest. Webstar with the newnitro plug-in, if you read this carefully jumps all over webten when usingOT, but as the report says, they are not done yet. And most of themagazine type articles are not doing real world testing. Not having readthe above articles, but most I have read do not test over dial up but overa 10-T lan...not real world to me.
And we've got two Tenon/WebTen- related notes:First, it took a while to get the relative importance of the various serverperformance parameters across to Tenon - they've been crowing aboutHits/Sec for a while now, and they weren't completely on top of issues likebandwidth, saturation, latency, or CPU load. But we spent a good amount oftime with them and hopefully we'll see them start to talk about the otherperformance parameters and benefits of WebTen (and not keep getting stuckon Hits/Sec).Second, we've been developing a new server testing/tuning tool that showslot more detailed information about the performance characteristics of theserver being tested and we're finding some interesting things.The big surprise (kind of) was that when both WebSTAR and WebTen arerunning under OpenTransport 1.3 (level playing field), WebSTAR isappreciably faster then WebTen. And with NITRO 2.0b1 installed, WebSTARwas about _twice_ as fast as WebTen for plain binary file hits.And switching to the Tenon TCP/IP stack speeds up _both_ servers. WebTengets more of a speed-up than WebSTAR with the Tenon stack, probably becauseof the low-level coupling of the Tenon ethernet driver and the WebTenengine. (And don't ask how we made WebSTAR use the Tenon TCP stack - itwasn't pretty and I wouldn't recommend it - but it DID make WebSTAR respondfaster, which was interesting.)To me, this is more data that suggests that OpenTransport 2.0 will closethe 'binary file hits per second' gap.We're going to try some testing of standardized SSI pages, too, which willeliminate any skew caused by low-level caching done by WebTen in the Tenonethernet driver.We'll publish everything (including the test tool!) when we're sure thateverything is solid, reproducable, and realistic. I think what we'llultimately see is that WebTen isn't quite as fast a 'server engine' asWebSTAR, but that the Tenon TCP stack more than makes up the difference.And OpenTransport 2.0 will be an interesting ingredient to add to _either_server!So, FireSite + WebTen is under technical investigation on both sides, andwe'll keep you posted.- -Mark===============================================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:
>Any comments about the recent press stating that WebTEN trounces WebSTAR>even at low volumes?>>I am using WebSTAR 3.0.... Should I sidegrade to WebTEN?>>http://www4.zdnet.com/pcmag/features/webserver98/bench.html>>http://macweek.zdnet.com/ns-search/1223/rv_web.html?NS-search-set=/35ad1/aaa>a000oqad1f5a&NS-docset=0&>Ahh but wait.. this is from todays firestite digest. Webstar with the newnitro plug-in, if you read this carefully jumps all over webten when usingOT, but as the report says, they are not done yet. And most of themagazine type articles are not doing real world testing. Not having readthe above articles, but most I have read do not test over dial up but overa 10-T lan...not real world to me.
And we've got two Tenon/WebTen- related notes:First, it took a while to get the relative importance of the various serverperformance parameters across to Tenon - they've been crowing aboutHits/Sec for a while now, and they weren't completely on top of issues likebandwidth, saturation, latency, or CPU load. But we spent a good amount oftime with them and hopefully we'll see them start to talk about the otherperformance parameters and benefits of WebTen (and not keep getting stuckon Hits/Sec).Second, we've been developing a new server testing/tuning tool that showslot more detailed information about the performance characteristics of theserver being tested and we're finding some interesting things.The big surprise (kind of) was that when both WebSTAR and WebTen arerunning under OpenTransport 1.3 (level playing field), WebSTAR isappreciably faster then WebTen. And with NITRO 2.0b1 installed, WebSTARwas about _twice_ as fast as WebTen for plain binary file hits.And switching to the Tenon TCP/IP stack speeds up _both_ servers. WebTengets more of a speed-up than WebSTAR with the Tenon stack, probably becauseof the low-level coupling of the Tenon ethernet driver and the WebTenengine. (And don't ask how we made WebSTAR use the Tenon TCP stack - itwasn't pretty and I wouldn't recommend it - but it DID make WebSTAR respondfaster, which was interesting.)To me, this is more data that suggests that OpenTransport 2.0 will closethe 'binary file hits per second' gap.We're going to try some testing of standardized SSI pages, too, which willeliminate any skew caused by low-level caching done by WebTen in the Tenonethernet driver.We'll publish everything (including the test tool!) when we're sure thateverything is solid, reproducable, and realistic. I think what we'llultimately see is that WebTen isn't quite as fast a 'server engine' asWebSTAR, but that the Tenon TCP stack more than makes up the difference.And OpenTransport 2.0 will be an interesting ingredient to add to _either_server!So, FireSite + WebTen is under technical investigation on both sides, andwe'll keep you posted.- -Mark===============================================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===============================================
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:
[WebDNA] [OT] hello (2012)
Letters as SKU (1998)
RE: type 2 errors with ssl server (1997)
Possible Bug in 2.0b15.acgi (1997)
Shorthand for command args (1998)
Can you set [index]=end+1 for [break]? (1998)
This message couldn't reach the list! (multi-column (1998)
Enhanced Master Counter? (1997)
Am on the list? (1997)
Nested search (1997)
WebCatalog Technical Reference (1997)
[append] and SSL (1997)
price totals (2003)
webdna dreamweaver extension. (2003)
webmerch and serials - almost there (1997)
Associative lookup style? + bit more (1997)
I give up!! (1997)
Re1000001: Setting up shop (1997)
calculating tax rates, mail order solutions and version 2 (1997)
Plugin or CGI or both (1997)