Re: [WebDNA] Not even sure what to ask for help on. . . :(
This WebDNA talk-list message is from 2008
It keeps the original formatting.
numero = 101639
interpreted = N
texte = David Bastedo wrote:> I stream through the same pipe and can handle up to 6mb a second -> which I have come close to, but not quite attained. The max I have hit> in the last week is 5.5.Pushing that close to the bandwidth limit is never a good idea. This may well be one cause of the problems but perhaps not the only one.---Rob wrote: > Which to me means that under normal conditions, low load, your lookups > are working fine, and everything functions normally. If it was a > problem in the coding, you would also see it in low load conditions as > well... yes/no?Not necessarily. An actual bug in the code would probably show up regardless but if it's a case of too complex a code for the server to handle, the scale of the problem would increase with the load.One "problem" with WebDNA is that it's a bit too useful sometimes. It's so tempting to keep adding another little script and yet another little script and...---Kenneth Grome wrote: > How many webdna hits are you getting per minute or second?That's a very important question although of course it also depends on the complexity of the code and the capacity of the server.A very rough guide:+ Less than 10 hits pr. minute: Something's really wrong if you have problems.+ 10-100 hits pr. minute: Should work fine unless the code is very complex and/or the server is old and slow.+ 100-500 hits pr. minute: Basic WebDNA functionality should work fine. Serious power required for advanced stuff.+ 500-1000 hits pr. minute: Time to seriously consider code optimisation and perhaps a faster server.+ More than 1000 hits pr. minute: You might want to consider a server cluster.---Ken's list of optimisation tips is a very good one (thanks Ken!) Here are a few more:+ Make sure you have enough *physical* RAM (virtual memory doesn't count) to run the software *and* cache all the templates and databases. This is by far the most important point. Nothing slows the server down more than reading from and writing to the harddisc.+ If you use temporary databases, keep them as simple as possible: e.g. just a a list of reference numbers pointing to a permanent database rather than all the data needed.+ Consider splitting large databases into smaller ones. But be careful: do this the wrong way and you end up increasing the load rather than reducing it.+ Consider storing large chunks of text as text files rather than in databases and use [Include] commands to fetch them. (Another method that may backfire so be careful.)Frank Nordberghttp://www.musicaviva.comhttp://stores.ebay.com/Nordbergs-Music-Store?refid=storehttp://www.abc-notation.comhttp://www.online-guitarist.comhttp://www.gitar-siden.comhttp://www.tablatvre.comhttp://www.mandolin-player.comhttp://www.blues-harmonica.comhttp://www.irish-banjo.comhttp://www.blues-banjo.comhttp://www.folk-banjo.comhttp://www.roarogfrank.com
Associated Messages, from the most recent to the oldest:
David Bastedo wrote:> I stream through the same pipe and can handle up to 6mb a second -> which I have come close to, but not quite attained. The max I have hit> in the last week is 5.5.Pushing that close to the bandwidth limit is never a good idea. This may well be one cause of the problems but perhaps not the only one.---Rob wrote: > Which to me means that under normal conditions, low load, your lookups > are working fine, and everything functions normally. If it was a > problem in the coding, you would also see it in low load conditions as > well... yes/no?Not necessarily. An actual bug in the code would probably show up regardless but if it's a case of too complex a code for the server to handle, the scale of the problem would increase with the load.One "problem" with WebDNA is that it's a bit too useful sometimes. It's so tempting to keep adding another little script and yet another little script and...---Kenneth Grome wrote: > How many webdna hits are you getting per minute or second?That's a very important question although of course it also depends on the complexity of the code and the capacity of the server.A very rough guide:+ Less than 10 hits pr. minute: Something's really wrong if you have problems.+ 10-100 hits pr. minute: Should work fine unless the code is very complex and/or the server is old and slow.+ 100-500 hits pr. minute: Basic WebDNA functionality should work fine. Serious power required for advanced stuff.+ 500-1000 hits pr. minute: Time to seriously consider code optimisation and perhaps a faster server.+ More than 1000 hits pr. minute: You might want to consider a server cluster.---Ken's list of optimisation tips is a very good one (thanks Ken!) Here are a few more:+ Make sure you have enough *physical* RAM (virtual memory doesn't count) to run the software *and* cache all the templates and databases. This is by far the most important point. Nothing slows the server down more than reading from and writing to the harddisc.+ If you use temporary databases, keep them as simple as possible: e.g. just a a list of reference numbers pointing to a permanent database rather than all the data needed.+ Consider splitting large databases into smaller ones. But be careful: do this the wrong way and you end up increasing the load rather than reducing it.+ Consider storing large chunks of text as text files rather than in databases and use
[include] commands to fetch them. (Another method that may backfire so be careful.)Frank Nordberghttp://www.musicaviva.comhttp://stores.ebay.com/Nordbergs-Music-Store?refid=storehttp://www.abc-notation.comhttp://www.online-guitarist.comhttp://www.gitar-siden.comhttp://www.tablatvre.comhttp://www.mandolin-player.comhttp://www.blues-harmonica.comhttp://www.irish-banjo.comhttp://www.blues-banjo.comhttp://www.folk-banjo.comhttp://www.roarogfrank.com
Frank Nordberg
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:
Setting Cart value (2000)
Ampersand Character (&) (1997)
UPDATE PROBLEM (1997)
[sendmail] on NT? (1997)
Projects & Contractors (1997)
[OT] DOD again *MYSTERY SOLVED* (2003)
Big Brother (just kidding, Ken) :) (2000)
[LISTFILES] (1998)
Merging databases (1997)
Frames and [cart] (1998)
[authenticate] was [TIME] not working (2007)
How To question on setting up downloads (1997)
carriage returns in data (1997)
SKU number. (2000)
WebCat2_Mac RETURNs in .db (1997)
PCS Frames (1997)
Summary search -- speed (1997)
Bug? (1997)
URGENT: NO BUG, woFIELDdata & FIELDword=ww (2002)
Version f1 status (1997)