Re: Here we go again...WebDNA - SQL- Clustering
This WebDNA talk-list message is from 2006
It keeps the original formatting.
numero = 66741
interpreted = N
texte = Oh boy...I like the last joke!-----Original Message-----From: WebDNA Talk [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf OfAlex McCombieSent: Tuesday, April 04, 2006 12:00 PMTo: WebDNA TalkSubject: Re: Here we go again...WebDNA - SQL- Clustering>> Again flat file has limitation on performance on server>=20> Bess, what you fail to mention is that a flat file database structure> can function just perfectly (with no limitations) for the majority> of needs out there. Sure, it depends on what you are targeting. WebDNA> does eCommerce, but it does LAN stuff and other functions as well. =These> blanket statements are just wrong.I don't know about majority because that too is debatable, but suffice =tosay that Donovan is right and that is works very well in numerous =projects.Like i have said in the past postings, we use them interchangeably =though Imust admit that I move to SQL tables instinctively now. The other thingmissing here is that WebDNA is first and foremost a markup language and =nota database system! And as a MUL it just plain rocks. My developers come =tothe table with php and .net experience and they LOVE, repeat LOVE DNA =forthe simplicity and small footprint the code makes. Very clean and quick =toget to the goal... MySQL, MSSQL or DNA flat files alike.=20>> Flat file makes it too much hassle to do any business intelligences,> data mining or even reporting. It is not scalable for enterprise =application.> Just flat out wrong. You are newbie to this list and to say webdna =can't> do reporting etc.. is just silly.Absolutely correct. Data mining and reporting are all easily done in DNAtables. She's right about the scalability issue though. I cant even tell =youhow many times the conversation about scaling multiple DNA installationsacross multiple servers has been hashed and rehashed on this list over =theyears by the best on the list. The truth is that is always fell short - =wayshort -- for us until we started fronting SQL with DNA. Now scalable is =justa given and WebDNA has been extended for us where we were facing thedecision to have to move away.=20>> Thanks Kenneth. Thanks for clarifying the myth.> Kenneth is a Myth... O.K., I just had to put that one in there for> enternainment> purposes.Now that's funny. I think even Ken would agree! :-)=20>> I think it is important for developers to understand the facts. =WebDNA is not>> "relational" database> Again, that's just plain wrong. It doesn't take a skilled developer to =do> this.> [search db=3D...&blah]> [founditems][lookup> =db=3Danother&value=3DID_OF_SECOND_DB_STORED_IN_FIRST_DATABASE&Blah]etc...=[/foundit> ems]> [/search]Its true that we build the relationships like Donovan mentioned above =but Ican tell you that I have found better performance (especially on large =datasets) with the join capability native to SQL than recursive searching =coulddo. On small datasets the recursive searching is ok but I have seen =outputswhere in order to show multiple columns of data 50 rows deep (which is =notunreasonable) with 5-6 columns of related data -- where you want to show =thetitles and not the related sku numbers... That led to recursive searchesthat number 300+ per page.Did it do it? Sure. Did it do it adequately on small data sets? You bet. =Butonce you try that on lets say a discussion forum, with 200,000 records =andtons of text, that recursive searches really starts to show its =limitations.This is just one real world (and not unreasonable) example where joinsearching on SQL is much faster. In fact I had to go back and =de-normalizemy data to put calculated totals into fields so as not to have to do thenumfound searching in a recursive lookup. It worked but it wasn't =elegant byany means.One other small point, though in the bigger context it is a smaller =point,by fronting DNA out to something like SQL you move the performance =suckingwork (searching) to another box or boxes dedicated to DB serving.Unfortunately with the built in RAM resident only access of DNA flat =files,it all has to happen on the same box as the webserver, which may or may =nothave it's own performance issues. Scalability is the huge winner in thiscase.=20>> Tab-delimited text files... It really makes many non-webdna =developers>> "wonder"> You make it sound like a text table is not ever used on the web. Think =Again!I got the same look all the time. Few of my clients came without their =ownIT dept. though. That being said, few wouldn't have some sort of =exposure tothings like SQL, SAP, etc. Flat files was something that we learned to =avoidin the conversations even though we kknew it would work for them. Now wesimply say everything is run from MSSQL and we can hit it with a variety =ofinterfaces. Our interface of choice is WebDNA because we can accomplish =thesame tasks in typically 1/2 the time and save you money.... Yet your =data isright there for any other future development/growth/reporting/etc. :-)Everyone wins. (including SMSI)>> Also once your flat file grow beyond roughly 2 GB (check with Network =Admin>> to verify the size),> This is a limit set by the OS and hard drive formatting... not webdna.>> you can't run data very well. The go-around solution is to break down =WebDNA>> database into smaller>> chunks for faster performance.>=20> Its not a "go-around" its just good programming, and in fact, is =called a> relational framework.Yes and no. Again, even with full normalization and excellent datastructuring skills you will still see data sets that large in the mostinnocent of places. I like to pick on discussion forums because I run =many.And at a couple hundred thousand records its starts to show it girth and =hasstability issues. Point is you cannot always employ workarounds, or =gooddatabase design to make your data 'smaller'. I have a hobby site that =gets 3million connections per month and 90% of that is discussion forum =activity.It has survived in DNA flat tables for years but is currently beingredeveloped in DNA/MSSQL. I expect to see great returns in performance,stability, expandability. BUT MY FRONT END IS STILL WEBDNA MARKUP =LANGUAGE!Because that's ultimately what WebDNA is a kick ass markup language.=20>> You really need Relationship Database for scaleable application.> Sigh. No you don't. You *may* need it.. but not always. It depends on =what the> client is doing.Definitely true.Just one man's opinion... Not trying to force it on anyone :-)Hope it helps to inform some and doesn't annoy others ;-)--=20Alex J McCombie =20Adventure Skies Interactive (ASI)Alex@Adventureskies.com-------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list
.To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to =Web Archive of this list is at: http://webdna.smithmicro.com/-------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list .To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://webdna.smithmicro.com/
Associated Messages, from the most recent to the oldest:
Oh boy...I like the last joke!-----Original Message-----From: WebDNA Talk [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf OfAlex McCombieSent: Tuesday, April 04, 2006 12:00 PMTo: WebDNA TalkSubject: Re: Here we go again...WebDNA - SQL- Clustering>> Again flat file has limitation on performance on server>=20> Bess, what you fail to mention is that a flat file database structure> can function just perfectly (with no limitations) for the majority> of needs out there. Sure, it depends on what you are targeting. WebDNA> does eCommerce, but it does LAN stuff and other functions as well. =These> blanket statements are just wrong.I don't know about majority because that too is debatable, but suffice =tosay that Donovan is right and that is works very well in numerous =projects.Like i have said in the past postings, we use them interchangeably =though Imust admit that I move to SQL tables instinctively now. The other thingmissing here is that WebDNA is first and foremost a markup language and =nota database system! And as a MUL it just plain rocks. My developers come =tothe table with php and .net experience and they LOVE, repeat LOVE DNA =forthe simplicity and small footprint the code makes. Very clean and quick =toget to the goal... MySQL, MSSQL or DNA flat files alike.=20>> Flat file makes it too much hassle to do any business intelligences,> data mining or even reporting. It is not scalable for enterprise =application.> Just flat out wrong. You are newbie to this list and to say webdna =can't> do reporting etc.. is just silly.Absolutely correct. Data mining and reporting are all easily done in DNAtables. She's right about the scalability issue though. I cant even tell =youhow many times the conversation about scaling multiple DNA installationsacross multiple servers has been hashed and rehashed on this list over =theyears by the best on the list. The truth is that is always fell short - =wayshort -- for us until we started fronting SQL with DNA. Now scalable is =justa given and WebDNA has been extended for us where we were facing thedecision to have to move away.=20>> Thanks Kenneth. Thanks for clarifying the myth.> Kenneth is a Myth... O.K., I just had to put that one in there for> enternainment> purposes.Now that's funny. I think even Ken would agree! :-)=20>> I think it is important for developers to understand the facts. =WebDNA is not>> "relational" database> Again, that's just plain wrong. It doesn't take a skilled developer to =do> this.> [search db=3D...&blah]> [founditems][lookup> =db=3Danother&value=3DID_OF_SECOND_DB_STORED_IN_FIRST_DATABASE&Blah]etc...=[/foundit> ems]> [/search]Its true that we build the relationships like Donovan mentioned above =but Ican tell you that I have found better performance (especially on large =datasets) with the join capability native to SQL than recursive searching =coulddo. On small datasets the recursive searching is ok but I have seen =outputswhere in order to show multiple columns of data 50 rows deep (which is =notunreasonable) with 5-6 columns of related data -- where you want to show =thetitles and not the related sku numbers... That led to recursive searchesthat number 300+ per page.Did it do it? Sure. Did it do it adequately on small data sets? You bet. =Butonce you try that on lets say a discussion forum, with 200,000 records =andtons of text, that recursive searches really starts to show its =limitations.This is just one real world (and not unreasonable) example where joinsearching on SQL is much faster. In fact I had to go back and =de-normalizemy data to put calculated totals into fields so as not to have to do thenumfound searching in a recursive lookup. It worked but it wasn't =elegant byany means.One other small point, though in the bigger context it is a smaller =point,by fronting DNA out to something like SQL you move the performance =suckingwork (searching) to another box or boxes dedicated to DB serving.Unfortunately with the built in RAM resident only access of DNA flat =files,it all has to happen on the same box as the webserver, which may or may =nothave it's own performance issues. Scalability is the huge winner in thiscase.=20>> Tab-delimited text files... It really makes many non-webdna =developers>> "wonder"> You make it sound like a text table is not ever used on the web. Think =Again!I got the same look all the time. Few of my clients came without their =ownIT dept. though. That being said, few wouldn't have some sort of =exposure tothings like SQL, SAP, etc. Flat files was something that we learned to =avoidin the conversations even though we kknew it would work for them. Now wesimply say everything is run from MSSQL and we can hit it with a variety =ofinterfaces. Our interface of choice is WebDNA because we can accomplish =thesame tasks in typically 1/2 the time and save you money.... Yet your =data isright there for any other future development/growth/reporting/etc. :-)Everyone wins. (including SMSI)>> Also once your flat file grow beyond roughly 2 GB (check with Network =Admin>> to verify the size),> This is a limit set by the OS and hard drive formatting... not webdna.>> you can't run data very well. The go-around solution is to break down =WebDNA>> database into smaller>> chunks for faster performance.>=20> Its not a "go-around" its just good programming, and in fact, is =called a> relational framework.Yes and no. Again, even with full normalization and excellent datastructuring skills you will still see data sets that large in the mostinnocent of places. I like to pick on discussion forums because I run =many.And at a couple hundred thousand records its starts to show it girth and =hasstability issues. Point is you cannot always employ workarounds, or =gooddatabase design to make your data 'smaller'. I have a hobby site that =gets 3million connections per month and 90% of that is discussion forum =activity.It has survived in DNA flat tables for years but is currently beingredeveloped in DNA/MSSQL. I expect to see great returns in performance,stability, expandability. BUT MY FRONT END IS STILL WEBDNA MARKUP =LANGUAGE!Because that's ultimately what WebDNA is a kick ass markup language.=20>> You really need Relationship Database for scaleable application.> Sigh. No you don't. You *may* need it.. but not always. It depends on =what the> client is doing.Definitely true.Just one man's opinion... Not trying to force it on anyone :-)Hope it helps to inform some and doesn't annoy others ;-)--=20Alex J McCombie =20Adventure Skies Interactive (ASI)Alex@Adventureskies.com-------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list .To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to =Web Archive of this list is at: http://webdna.smithmicro.com/-------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list .To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://webdna.smithmicro.com/
"Bess Ho"
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:
Summ (1998)
Setting variables. (1998)
[isfile] ? (1997)
looking for developers with solutions (2000)
WebCat b13 CGI -shownext- (1997)
WebCatalog for Postcards ? (1997)
RE: Questions To Answer (1997)
Perl programmer needed (2002)
New Command prefs ... (1997)
[WebDNA] How do we tell what's misconfigured? Or which WebDNA 7 version to use? (2011)
why am I getting an authenticate dialog with no [protect]? (2000)
WebCat2b13MacPlugIn - [showif][search][/showif] (1997)
Date AND time (2001)
OK, here goes... (1997)
WC1.6 to WC2 date formatting (1997)
verify email address (1998)
Stopping bad HTML propagation ? (1997)
[WebDNA] Conversion of 500.000 records; advise needed (2008)
Single Link browsing (1997)
Linux file configuration (1999)