RE: Re[2]: sequence numbers
We were asked, not too long ago, to create one Oracle8i database with only *one* table with some 700+ columns. While at it, the consultant (hired by end user dept) also suggested that we make it an IOT using an LMT, and since the table will never grow over 1GB, asked if there was a way to put it in KEEP buffer pool. He was helping re-write/enhance some MS Access Apps. Talk about knowing all the right lingo... ;) - Kirti -Original Message- Sent: Friday, October 11, 2002 8:59 AM To: Multiple recipients of list ORACLE-L April, What can I say? Ouch! I feel your pain. I've been trapped in some pretty ridiculous situations too. (Though, I think you have me beat! A 37 column primary key?? Really??) Well, you at least seem to have the proper attitude. ;-) Without a sense of humor, I'm afraid you'd go insane in short order! ;-) The only other thing I can think of when people shut you down like that is: document. At meeting X, on such and such a date, I identified this problem, and Mr. Z told me to not to worry about it. It may not help, but from a sanity point of view, there is a certain amount of satisfaction in I told you so!, even if you never verbalize it;-) Hang in there, -Mark On Fri, 2002-10-11 at 08:43, April Wells wrote: Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but thinking makes it so. -Shakespeare !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0 Transitional//EN HTMLHEAD META content=text/html; charset=iso-8859-1 http-equiv=Content-Type META content=MSHTML 5.00.2314.1000 name=GENERATOR STYLE/STYLE /HEAD BODY
RE: Re[2]: sequence numbers
it's all in the buzzwords, obviously :) --- Deshpande, Kirti [EMAIL PROTECTED] wrote: We were asked, not too long ago, to create one Oracle8i database with only *one* table with some 700+ columns. While at it, the consultant (hired by end user dept) also suggested that we make it an IOT using an LMT, and since the table will never grow over 1GB, asked if there was a way to put it in KEEP buffer pool. He was helping re-write/enhance some MS Access Apps. Talk about knowing all the right lingo... ;) - Kirti -Original Message- Sent: Friday, October 11, 2002 8:59 AM To: Multiple recipients of list ORACLE-L April, What can I say? Ouch! I feel your pain. I've been trapped in some pretty ridiculous situations too. (Though, I think you have me beat! A 37 column primary key?? Really??) Well, you at least seem to have the proper attitude. ;-) Without a sense of humor, I'm afraid you'd go insane in short order! ;-) The only other thing I can think of when people shut you down like that is: document. At meeting X, on such and such a date, I identified this problem, and Mr. Z told me to not to worry about it. It may not help, but from a sanity point of view, there is a certain amount of satisfaction in I told you so!, even if you never verbalize it;-) Hang in there, -Mark On Fri, 2002-10-11 at 08:43, April Wells wrote: Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but
RE: Re[2]: sequence numbers
But the DOCUMENTATION says
8-0
April Wells
Oracle DBA
Keep yourself well oiled with life, laughter, new ideas and action.
Otherwise you will rust out. _Anonymous
-Original Message-
Sent: Friday, October 11, 2002 10:20 AM
To: Multiple recipients of list ORACLE-L
it's all in the buzzwords, obviously :)
--- Deshpande, Kirti [EMAIL PROTECTED] wrote:
We were asked, not too long ago, to create one Oracle8i database with
only
*one* table with some 700+ columns. While at it, the consultant
(hired by
end user dept) also suggested that we make it an IOT using an LMT,
and since
the table will never grow over 1GB, asked if there was a way to put
it in
KEEP buffer pool. He was helping re-write/enhance some MS Access
Apps.
Talk about knowing all the right lingo... ;)
- Kirti
begin 666 InterScan_Disclaimer.txt
M5AE(EN9F]R;6%T:6]N(-O;G1A:6YE9!I;B!T:ES(4M;6%I;!IR!S
M=')I8W1L2!C;VYF:61E;G1I86P@86YD(9OB!T:4@:6YT96YD960@=7-E
M(]F('1H92!A91R97-S964@;VYL3L@:70@;6%Y(%LV\@8F4@;5G86QL
M2!PFEV:6QE9V5D(%N9]OB!PFEC92!S96YS:71I=F4N(!.;W1I8V4@
M:7,@:5R96)Y(=I=F5N('1H870@86YY(1IV-L;W-UF4L('5S92!OB!C
M;W!Y:6YG(]F('1H92!I;F9OFUA=EO;B!B2!A;GEO;F4@;W1H97(@=AA
M;B!T:4@:6YT96YD960@F5C:7!I96YT(ES('!R;VAI8FET960@86YD(UA
M2!B92!I;QE9V%L+B @268@6]U(AA=F4@F5C96EV960@=AIR!M97-S
M86=E(EN(5RF]R+!P;5AV4@;F]T:69Y('1H92!S96YD97(@:6UM961I
M871E;'D@8GD@F5T=7)N(4M;6%I;X*D-OG!OF%T92!37-T96US+!)
M;F,N(AAR!T86ME;B!E=F5R2!R96%S;VYA8FQE('!R96-A=71I;VX@=\@
M96YS=7)E('1H870@86YY(%T=%C:UE;G0@=\@=AIR!E+6UA:6P@:%S
M()E96X@W=E'0@9F]R('9IG5S97,N(!792!A8V-E'0@;F\@;EA8FEL
M:71Y(9OB!A;GD@9%M86=E('-UW1A:6YE9!AR!A(')EW5L=!O9B!S
M;V9T=V%R92!V:7)UV5S(%N9!A9'9IV4@6]U(-AG)Y(]U=!Y;W5R
M(]W;B!V:7)UR!C:5C:W,@8F5F;W)E(]P96YI;F@86YY(%T=%C:UE
+;G0N#0H-@T*#0H
end
--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: April Wells
INET: [EMAIL PROTECTED]
Fat City Network Services-- 858-538-5051 http://www.fatcity.com
San Diego, California-- Mailing list and web hosting services
-
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from). You may
also send the HELP command for other information (like subscribing).
Re:RE: Re[2]: sequence numbers
Let's see, 1 table with 700+ columns that can grow to ~1GB that you want to iot and have in the keep pool. What are you smoking! That's one consultant that I'd HAVE to laugh in his/her face. And he/she would NOT get away with it. Dick Goulet Reply Separator Author: Rachel Carmichael [EMAIL PROTECTED] Date: 10/11/2002 7:19 AM it's all in the buzzwords, obviously :) --- Deshpande, Kirti [EMAIL PROTECTED] wrote: We were asked, not too long ago, to create one Oracle8i database with only *one* table with some 700+ columns. While at it, the consultant (hired by end user dept) also suggested that we make it an IOT using an LMT, and since the table will never grow over 1GB, asked if there was a way to put it in KEEP buffer pool. He was helping re-write/enhance some MS Access Apps. Talk about knowing all the right lingo... ;) - Kirti -Original Message- Sent: Friday, October 11, 2002 8:59 AM To: Multiple recipients of list ORACLE-L April, What can I say? Ouch! I feel your pain. I've been trapped in some pretty ridiculous situations too. (Though, I think you have me beat! A 37 column primary key?? Really??) Well, you at least seem to have the proper attitude. ;-) Without a sense of humor, I'm afraid you'd go insane in short order! ;-) The only other thing I can think of when people shut you down like that is: document. At meeting X, on such and such a date, I identified this problem, and Mr. Z told me to not to worry about it. It may not help, but from a sanity point of view, there is a certain amount of satisfaction in I told you so!, even if you never verbalize it;-) Hang in there, -Mark On Fri, 2002-10-11 at 08:43, April Wells wrote: Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L
RE: RE: Re[2]: sequence numbers
I wanted to take a picture with him. .. and take him out to lunch to learn from his experience ... ;-) but it turned out he lasted only for less than a week... ;) (Some developers he was working with knew a bit more Oracle than him) - Kirti -Original Message- Sent: Friday, October 11, 2002 11:49 AM To: Multiple recipients of list ORACLE-L Let's see, 1 table with 700+ columns that can grow to ~1GB that you want to iot and have in the keep pool. What are you smoking! That's one consultant that I'd HAVE to laugh in his/her face. And he/she would NOT get away with it. Dick Goulet Reply Separator Author: Rachel Carmichael [EMAIL PROTECTED] Date: 10/11/2002 7:19 AM it's all in the buzzwords, obviously :) --- Deshpande, Kirti [EMAIL PROTECTED] wrote: We were asked, not too long ago, to create one Oracle8i database with only *one* table with some 700+ columns. While at it, the consultant (hired by end user dept) also suggested that we make it an IOT using an LMT, and since the table will never grow over 1GB, asked if there was a way to put it in KEEP buffer pool. He was helping re-write/enhance some MS Access Apps. Talk about knowing all the right lingo... ;) - Kirti -Original Message- Sent: Friday, October 11, 2002 8:59 AM To: Multiple recipients of list ORACLE-L April, What can I say? Ouch! I feel your pain. I've been trapped in some pretty ridiculous situations too. (Though, I think you have me beat! A 37 column primary key?? Really??) Well, you at least seem to have the proper attitude. ;-) Without a sense of humor, I'm afraid you'd go insane in short order! ;-) The only other thing I can think of when people shut you down like that is: document. At meeting X, on such and such a date, I identified this problem, and Mr. Z told me to not to worry about it. It may not help, but from a sanity point of view, there is a certain amount of satisfaction in I told you so!, even if you never verbalize it;-) Hang in there, -Mark On Fri, 2002-10-11 at 08:43, April Wells wrote: Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re:
RE: Re[2]: sequence numbers
Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but thinking makes it so. -Shakespeare !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0 Transitional//EN HTMLHEAD META content=text/html; charset=iso-8859-1 http-equiv=Content-Type META content=MSHTML 5.00.2314.1000 name=GENERATOR STYLE/STYLE /HEAD BODY bgColor=#ff style=FONT: 10pt Times New Roman; MARGIN-LEFT: 2px; MARGIN-TOP: 2px DIVFONT face=Arial size=3I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval./FONT/DIV DIVnbsp;/DIV DIVYechiel AdarBRMehish/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: blackBFrom:/B A href=mailto:Tim;SageLogix.com [EMAIL PROTECTED]Tim Gorman/A /DIV DIV style=FONT: 10pt arialBTo:/B A href=mailto:ORACLE-L;fatcity.com [EMAIL PROTECTED]Multiple recipients of list ORACLE-L/A /DIV DIV style=FONT: 10pt arialBSent:/B Thursday, October 10, 2002 7:43 PM/DIV DIV style=FONT: 10pt arialBSubject:/B Re: sequence numbers/DIV DIVBR/DIV DIVFONT face=ArialCACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else.../FONT/DIV DIVFONT face=Arial/FONTnbsp;/DIV DIVFONT face=Arial...of course, I get the feeling that that wasn't the gist of your question, was it?/FONT/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px;
RE: Re[2]: sequence numbers
April, What can I say? Ouch! I feel your pain. I've been trapped in some pretty ridiculous situations too. (Though, I think you have me beat! A 37 column primary key?? Really??) Well, you at least seem to have the proper attitude. ;-) Without a sense of humor, I'm afraid you'd go insane in short order! ;-) The only other thing I can think of when people shut you down like that is: document. At meeting X, on such and such a date, I identified this problem, and Mr. Z told me to not to worry about it. It may not help, but from a sanity point of view, there is a certain amount of satisfaction in I told you so!, even if you never verbalize it;-) Hang in there, -Mark On Fri, 2002-10-11 at 08:43, April Wells wrote: Mark... If this were the MOST serious design flaw in the whole mess, I wouldn't care so much. There is a point where you just shut up (gee, I have been TOLD to do that in meetings) and wait till it breaks (or worse, one of our clients buys it and we have to TRY to implement). I am the funny one... the one to laugh at and make fun of because I keep trying to tell them that you can't do things. You can't have a totally denormalized Oracle table if there 1500 columns in it... yes queries will fly on a table that can't be built. You can't have 37 columns in a primary key. Date really isn't an acceptable name for a column. April Wells Oracle DBA Keep yourself well oiled with life, laughter, new ideas and action. Otherwise you will rust out. _Anonymous -Original Message- Sent: Thursday, October 10, 2002 7:34 PM To: Multiple recipients of list ORACLE-L Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but thinking makes it so. -Shakespeare !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0 Transitional//EN HTMLHEAD META content=text/html; charset=iso-8859-1 http-equiv=Content-Type META content=MSHTML 5.00.2314.1000 name=GENERATOR STYLE/STYLE /HEAD BODY bgColor=#ff style=FONT: 10pt Times New Roman; MARGIN-LEFT: 2px; MARGIN-TOP: 2px DIVFONT face=Arial size=3I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval./FONT/DIV DIVnbsp;/DIV DIVYechiel AdarBRMehish/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND:
Re[2]: sequence numbers
Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but thinking makes it so. -Shakespeare !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0 Transitional//EN HTMLHEAD META content=text/html; charset=iso-8859-1 http-equiv=Content-Type META content=MSHTML 5.00.2314.1000 name=GENERATOR STYLE/STYLE /HEAD BODY bgColor=#ff style=FONT: 10pt Times New Roman; MARGIN-LEFT: 2px; MARGIN-TOP: 2px DIVFONT face=Arial size=3I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval./FONT/DIV DIVnbsp;/DIV DIVYechiel AdarBRMehish/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: blackBFrom:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Tim Gorman/A /DIV DIV style=FONT: 10pt arialBTo:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Multiple recipients of list ORACLE-L/A /DIV DIV style=FONT: 10pt arialBSent:/B Thursday, October 10, 2002 7:43 PM/DIV DIV style=FONT: 10pt arialBSubject:/B Re: sequence numbers/DIV DIVBR/DIV DIVFONT face=ArialCACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else.../FONT/DIV DIVFONT face=Arial/FONTnbsp;/DIV DIVFONT face=Arial...of course, I get the feeling that that wasn't the gist of your question, was it?/FONT/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: blackBFrom:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]April Wells/A /DIV DIV style=FONT: 10pt arialBTo:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Multiple recipients of list ORACLE-L/A /DIV DIV style=FONT: 10pt arialBSent:/B Wednesday, October 09, 2002 8:54 AM/DIV DIV style=FONT: 10pt arialBSubject:/B sequence numbers/DIV DIVBR/DIV DIVSPAN class=841194713-09102002I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads.nbsp; How huge is the potential performancenbsp;hit if I take out the cache 20?/SPAN/DIV DIVnbsp;/DIV PFONT face=Courier NewApril Wells/FONT BRFONT face=Courier NewOracle DBAnbsp;/FONTBRSPAN class=841194713-09102002FONT face=Courier NewTSPAN class=841194713-09102002here is neither good nor bad, but thinking makes it so. -Shakespeare/SPAN/FONT/SPAN/P/BLOCKQUOTE/BLOCKQUOTE/BODY/HTML -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- 858-538-5051 http://www.fatcity.com San Diego, California-- Mailing list and web hosting services - To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: Re[2]: sequence numbers
Hi Dick, I have to disagree with you here. Particularly in the case where this sequence will see any sort of concurrency, from multiple concurrent sessions accessing it. This is due to the serialization on the SQ enqueue. This will cause far worse scalability issues than any I/O. Not that I/O is insignificant, but in this situation, serialization on the enqueue will be the real showstopper for scalability. As to losing the cached values, well, so what? If your design is such that it's important to have an unbroken contiguous sequence of numbers with no gaps, then I would argue that is a serious design flaw. Also, if that's your requirement, then a sequence is not appropriate, since it can and will end up causing gaps, the first time you roll back a transaction. Finally, as to sequences losing cached values, unless your instance crashes or does a shutdown abort, Oracle should not loose any sequence values. -Mark On Thu, 2002-10-10 at 18:18, [EMAIL PROTECTED] wrote: Actually there is no IO penalty since Oracle has to treat the sequence just like any table with the old LRU algorithm. I have several sequences with a cache of 0 and they perform as well as those with a cache value. The big difference is when you shut down the database and all of those cached values end up in the trash. Dick Goulet Reply Separator Author: Yechiel Adar [EMAIL PROTECTED] Date: 10/10/2002 1:38 PM I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval. Yechiel Adar Mehish - Original Message - From: Tim Gorman To: Multiple recipients of list ORACLE-L Sent: Thursday, October 10, 2002 7:43 PM Subject: Re: sequence numbers CACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else... ...of course, I get the feeling that that wasn't the gist of your question, was it? - Original Message - From: April Wells To: Multiple recipients of list ORACLE-L Sent: Wednesday, October 09, 2002 8:54 AM Subject: sequence numbers I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads. How huge is the potential performance hit if I take out the cache 20? April Wells Oracle DBA There is neither good nor bad, but thinking makes it so. -Shakespeare !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0 Transitional//EN HTMLHEAD META content=text/html; charset=iso-8859-1 http-equiv=Content-Type META content=MSHTML 5.00.2314.1000 name=GENERATOR STYLE/STYLE /HEAD BODY bgColor=#ff style=FONT: 10pt Times New Roman; MARGIN-LEFT: 2px; MARGIN-TOP: 2px DIVFONT face=Arial size=3I think that you will have an update to the sequence number EVERY time instead of every 20 times. That's mean I/o for every nextval./FONT/DIV DIVnbsp;/DIV DIVYechiel AdarBRMehish/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: blackBFrom:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Tim Gorman/A /DIV DIV style=FONT: 10pt arialBTo:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Multiple recipients of list ORACLE-L/A /DIV DIV style=FONT: 10pt arialBSent:/B Thursday, October 10, 2002 7:43 PM/DIV DIV style=FONT: 10pt arialBSubject:/B Re: sequence numbers/DIV DIVBR/DIV DIVFONT face=ArialCACHE 20 is the default, so if you remove the clause, it will have absolutely no impact on performance or anything else.../FONT/DIV DIVFONT face=Arial/FONTnbsp;/DIV DIVFONT face=Arial...of course, I get the feeling that that wasn't the gist of your question, was it?/FONT/DIV BLOCKQUOTE dir=ltr style=BORDER-LEFT: #00 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: 0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px DIV style=FONT: 10pt arial- Original Message - /DIV DIV style=BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: blackBFrom:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]April Wells/A /DIV DIV style=FONT: 10pt arialBTo:/B A href=mailto:[EMAIL PROTECTED]; [EMAIL PROTECTED]Multiple recipients of list ORACLE-L/A /DIV DIV style=FONT: 10pt arialBSent:/B Wednesday, October 09, 2002 8:54 AM/DIV DIV style=FONT: 10pt arialBSubject:/B sequence numbers/DIV DIVBR/DIV DIVSPAN class=841194713-09102002I have been given create scripts for sequences to be used in tables that will be loaded via bulk loads.nbsp; How huge is the potential performancenbsp;hit if I take out the cache 20?/SPAN/DIV DIVnbsp;/DIV PFONT face=Courier NewApril Wells/FONT BRFONT
