ESQL/C -Insert- Performance

ESQL/C -Insert- Performance

Post by Inge Dille » Sun, 31 Dec 1899 09:00:00



Hi all,

Does anyone know some tricks to improve the performance of an insert into a big
table.

Here is some info:
We run Informix Dynamic Server Version 7.31.UC3
The table has currently 8214357 rows in it.
Row size=265
The table is fragmented.
Indexes:
Index name          Owner     Type    Cluster  Columns

pstidx_1            informix  dupls   No       a_sub_num

pstidx_2            informix  dupls   No       a_sub_imsi

pstidx_3            informix  dupls   No       a_sub_imei

pstidx_4            informix  dupls   No       b_sub_num
                                                                b_sub_imsi

b_sub_imei

pstidx_5            informix  dupls   No       ch_start_time

A C-program runs every two hours. At each run about 75000 records are inserted
into the table. This amount will increase significantly in future.
The C-program gets a file_name passed through by a Unix shell script. It
connects to the database and checks the file name in a table with only 2 fields
(file_name and status). If it finds the file_name the data from that file is not
been loaded again. If it is a new file the file_name is inserted in the
file_name table. Each run loads data from 32 files.
The data from the flatfile is read into a C-structure. The MAXSIZE of the
structure is set to 10000.
The ESQL/C variables are filled from the structure.
A statement is prepared:
 sprintf(insert_query," INSERT INTO %s VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

    EXEC SQL PREPARE insert_cdr FROM :insert_query;
    EXEC SQL EXECUTE insert_cdr USING
                                :record_type,
                                :a_sub_num,
                                :a_sub_imsi,
                                ................
When the amount of records > MAXSIZE the program stays in the loop until it
encounters EOF .

Before the indexes were created the program took about 10 minutes to run. With
the indexes it runs to slow to be acceptable.

Any suggestions are welcome!

Thanks,

Inge Dillen

 
 
 

ESQL/C -Insert- Performance

Post by Obnoxio The Clow » Sun, 31 Dec 1899 09:00:00



Quote:

>Does anyone know some tricks to improve the performance of an insert into a
>big
>table.

>Here is some info:
>We run Informix Dynamic Server Version 7.31.UC3
>The table has currently 8214357 rows in it.
>Row size=265
>The table is fragmented.
>Indexes:
>Index name          Owner     Type    Cluster  Columns

>pstidx_1            informix  dupls   No       a_sub_num

>pstidx_2            informix  dupls   No       a_sub_imsi

>pstidx_3            informix  dupls   No       a_sub_imei

>pstidx_4            informix  dupls   No       b_sub_num
>                                                                 b_sub_imsi

>b_sub_imei

>pstidx_5            informix  dupls   No       ch_start_time

I think these are too many indexes for the loading requirements, but how
about using onpload? Have your C program generate an unload file and use
onpload to load the generated file?

- Show quoted text -

Quote:>A C-program runs every two hours. At each run about 75000 records are
>inserted
>into the table. This amount will increase significantly in future.
>The C-program gets a file_name passed through by a Unix shell script. It
>connects to the database and checks the file name in a table with only 2
>fields
>(file_name and status). If it finds the file_name the data from that file
>is not
>been loaded again. If it is a new file the file_name is inserted in the
>file_name table. Each run loads data from 32 files.
>The data from the flatfile is read into a C-structure. The MAXSIZE of the
>structure is set to 10000.
>The ESQL/C variables are filled from the structure.
>A statement is prepared:
>  sprintf(insert_query," INSERT INTO %s
>VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
>,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

>     EXEC SQL PREPARE insert_cdr FROM :insert_query;
>     EXEC SQL EXECUTE insert_cdr USING
>                                 :record_type,
>                                 :a_sub_num,
>                                 :a_sub_imsi,
>                                 ................
>When the amount of records > MAXSIZE the program stays in the loop until it
>encounters EOF .

>Before the indexes were created the program took about 10 minutes to run.
>With
>the indexes it runs to slow to be acceptable.

Drop the indexes and create them just before they're needed again?
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

 
 
 

ESQL/C -Insert- Performance

Post by mars1.. » Sun, 31 Dec 1899 09:00:00


Use an insert cursor.

exec sql prepare ins_stat from 'insert into...';
exec sql declare ins_rec cursor for ins_stat;
exec sql open ins_rec;
exec sql put ins_rec using ...;
every hundred records or so (you'll have to do some research on what
works best for you),
exec sql flush ins_rec.

That should inprove performance by at least 25% over the execute you
have now.



> Hi all,

> Does anyone know some tricks to improve the performance of an insert
into a big
> table.

> Here is some info:
> We run Informix Dynamic Server Version 7.31.UC3
> The table has currently 8214357 rows in it.
> Row size=265
> The table is fragmented.
> Indexes:
> Index name Owner Type Cluster Columns

> pstidx_1 informix dupls No a_sub_num

> pstidx_2 informix dupls No a_sub_imsi

> pstidx_3 informix dupls No a_sub_imei

> pstidx_4 informix dupls No b_sub_num
> b_sub_imsi

> b_sub_imei

> pstidx_5 informix dupls No ch_start_time

> A C-program runs every two hours. At each run about 75000 records are
inserted
> into the table. This amount will increase significantly in future.
> The C-program gets a file_name passed through by a Unix shell script.
It
> connects to the database and checks the file name in a table with only
2 fields
> (file_name and status). If it finds the file_name the data from that
file is not
> been loaded again. If it is a new file the file_name is inserted in
the
> file_name table. Each run loads data from 32 files.
> The data from the flatfile is read into a C-structure. The MAXSIZE of
the
> structure is set to 10000.
> The ESQL/C variables are filled from the structure.
> A statement is prepared:
> sprintf(insert_query," INSERT INTO %s

VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?

- Show quoted text -

> ,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

> EXEC SQL PREPARE insert_cdr FROM :insert_query;
> EXEC SQL EXECUTE insert_cdr USING
> :record_type,
> :a_sub_num,
> :a_sub_imsi,
> ................
> When the amount of records > MAXSIZE the program stays in the loop
until it
> encounters EOF .

> Before the indexes were created the program took about 10 minutes to
run. With
> the indexes it runs to slow to be acceptable.

> Any suggestions are welcome!

> Thanks,

> Inge Dillen


--
# unrm /
ksh: unrm: not found
# man cpio

Sent via Deja.com http://www.deja.com/
Before you buy.

 
 
 

ESQL/C -Insert- Performance

Post by Rekaish Bhardwa » Sun, 31 Dec 1899 09:00:00


Try dropping some or all of your indexes and then recreate them after the
inserts have finished.
You may find that this is quicker. You may also have to run an update
statistics on your table
after you've re-created your indexes.

Cheers
Rekaish

-----Original Message-----

Sent: 24 March 2000 10:25

Subject: ESQL/C -Insert- Performance

Hi all,

Does anyone know some tricks to improve the performance of an insert into a
big
table.

Here is some info:
We run Informix Dynamic Server Version 7.31.UC3
The table has currently 8214357 rows in it.
Row size=265
The table is fragmented.
Indexes:
Index name          Owner     Type    Cluster  Columns

pstidx_1            informix  dupls   No       a_sub_num

pstidx_2            informix  dupls   No       a_sub_imsi

pstidx_3            informix  dupls   No       a_sub_imei

pstidx_4            informix  dupls   No       b_sub_num
                                                                b_sub_imsi

b_sub_imei

pstidx_5            informix  dupls   No       ch_start_time

A C-program runs every two hours. At each run about 75000 records are
inserted
into the table. This amount will increase significantly in future.
The C-program gets a file_name passed through by a Unix shell script. It
connects to the database and checks the file name in a table with only 2
fields
(file_name and status). If it finds the file_name the data from that file is
not
been loaded again. If it is a new file the file_name is inserted in the
file_name table. Each run loads data from 32 files.
The data from the flatfile is read into a C-structure. The MAXSIZE of the
structure is set to 10000.
The ESQL/C variables are filled from the structure.
A statement is prepared:
 sprintf(insert_query," INSERT INTO %s
VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

    EXEC SQL PREPARE insert_cdr FROM :insert_query;
    EXEC SQL EXECUTE insert_cdr USING
                                :record_type,
                                :a_sub_num,
                                :a_sub_imsi,
                                ................
When the amount of records > MAXSIZE the program stays in the loop until it
encounters EOF .

Before the indexes were created the program took about 10 minutes to run.
With
the indexes it runs to slow to be acceptable.

Any suggestions are welcome!

Thanks,

Inge Dillen


 
 
 

ESQL/C -Insert- Performance

Post by a_blo.. » Sun, 31 Dec 1899 09:00:00



> Hi all,

> Does anyone know some tricks to improve the performance of an insert into a big
> table.

> Here is some info:
> We run Informix Dynamic Server Version 7.31.UC3
> The table has currently 8214357 rows in it.
> Row size=265
> The table is fragmented.
> Indexes:
> Index name          Owner     Type    Cluster  Columns

> pstidx_1            informix  dupls   No       a_sub_num

> pstidx_2            informix  dupls   No       a_sub_imsi

> pstidx_3            informix  dupls   No       a_sub_imei

> pstidx_4            informix  dupls   No       b_sub_num
>                                                                 b_sub_imsi

> b_sub_imei

> pstidx_5            informix  dupls   No       ch_start_time

> A C-program runs every two hours. At each run about 75000 records are inserted
> into the table. This amount will increase significantly in future.
> The C-program gets a file_name passed through by a Unix shell script. It
> connects to the database and checks the file name in a table with only 2 fields
> (file_name and status). If it finds the file_name the data from that file is not
> been loaded again. If it is a new file the file_name is inserted in the
> file_name table. Each run loads data from 32 files.
> The data from the flatfile is read into a C-structure. The MAXSIZE of the
> structure is set to 10000.
> The ESQL/C variables are filled from the structure.
> A statement is prepared:
>  sprintf(insert_query," INSERT INTO %s VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
> ,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

>     EXEC SQL PREPARE insert_cdr FROM :insert_query;
>     EXEC SQL EXECUTE insert_cdr USING
>                                 :record_type,
>                                 :a_sub_num,
>                                 :a_sub_imsi,
>                                 ................
> When the amount of records > MAXSIZE the program stays in the loop until it
> encounters EOF .

> Before the indexes were created the program took about 10 minutes to run. With
> the indexes it runs to slow to be acceptable.

Take a look at insert cursors Dude. Also, make sure you don't prepare the insert inside that loop.

Did you put the indexes in another dbspace and not allow them to be fragmented?

You could also run a different program for each fragment so that you can insert in parallel. At least that is what I would do in 4GL, but perhaps you can use multiple threads in ESQL?

AB
----------------------------------------------------------------
Get your free email from AltaVista at http://altavista.iname.com

 
 
 

ESQL/C -Insert- Performance

Post by a_blo.. » Sun, 31 Dec 1899 09:00:00



> Try dropping some or all of your indexes and then recreate them after the
> inserts have finished.
> You may find that this is quicker. You may also have to run an update
> statistics on your table
> after you've re-created your indexes.

I am not too clear on this Dude. Are you saying that every 2 hours, they should drop all indexes, load 75000 records, then create 5 indexes on over 8 million records, and all in under 2 hours?

> -----Original Message-----

> Sent: 24 March 2000 10:25

> Subject: ESQL/C -Insert- Performance

> Hi all,

> Does anyone know some tricks to improve the performance of an insert into a
> big
> table.

> Here is some info:
> We run Informix Dynamic Server Version 7.31.UC3
> The table has currently 8214357 rows in it.
> Row size=265
> The table is fragmented.
> Indexes:
> Index name          Owner     Type    Cluster  Columns

> pstidx_1            informix  dupls   No       a_sub_num

> pstidx_2            informix  dupls   No       a_sub_imsi

> pstidx_3            informix  dupls   No       a_sub_imei

> pstidx_4            informix  dupls   No       b_sub_num
>                                                                 b_sub_imsi

> b_sub_imei

> pstidx_5            informix  dupls   No       ch_start_time

> A C-program runs every two hours. At each run about 75000 records are
> inserted
> into the table. This amount will increase significantly in future.
> The C-program gets a file_name passed through by a Unix shell script. It
> connects to the database and checks the file name in a table with only 2
> fields
> (file_name and status). If it finds the file_name the data from that file is
> not
> been loaded again. If it is a new file the file_name is inserted in the
> file_name table. Each run loads data from 32 files.
> The data from the flatfile is read into a C-structure. The MAXSIZE of the
> structure is set to 10000.
> The ESQL/C variables are filled from the structure.
> A statement is prepared:
>  sprintf(insert_query," INSERT INTO %s
> VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
> ,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

>     EXEC SQL PREPARE insert_cdr FROM :insert_query;
>     EXEC SQL EXECUTE insert_cdr USING
>                                 :record_type,
>                                 :a_sub_num,
>                                 :a_sub_imsi,
>                                 ................
> When the amount of records > MAXSIZE the program stays in the loop until it
> encounters EOF .

> Before the indexes were created the program took about 10 minutes to run.
> With
> the indexes it runs to slow to be acceptable.

> Any suggestions are welcome!

> Thanks,

> Inge Dillen


AB
----------------------------------------------------------------
Get your free email from AltaVista at http://altavista.iname.com
 
 
 

ESQL/C -Insert- Performance

Post by Hannes Visagi » Sun, 31 Dec 1899 09:00:00


Hi

Run your loads in parallel, check PDQ, or

One way to get your program to load faster is to let it fork say +-10
children that will do the inserts.

Your Parent program will break up {calculate} the load file into 10
begin/end bits.  Now fork 10 children passing the start stop values as
parameters to the children.  Your program will run +- 10 times faster.

I used to load big tables like this with Online 5 {Before HPL}, but be
aware that you will use just about 100% of all the resources on a 4 CPU
server.  You can play with the number of child's to fork.  This is the best
way to load data fast on a multi CPU server with Online 7.  I am not to
sure if you will get the same improvement with IDS 7x+

I Have the source code {SCO UNIXWARE SYS V} available at my previous
Client.  If you need it, email me.  Hope they still have it, but not to
difficult to reproduce once U know how write code to fork a child in UNIX.
 There is some extra compiler switches to use when U do multi threading in
ESQLC.

Hope It helps.

Hannes Visagie
Systems Engineer
MIBCO

-----Original Message-----

Sent:   Friday, March 24, 2000 12:25 PM

Subject:        ESQL/C -Insert- Performance

Hi all,

Does anyone know some tricks to improve the performance of an insert into a
big
table.

Here is some info:
We run Informix Dynamic Server Version 7.31.UC3
The table has currently 8214357 rows in it.
Row size=265
The table is fragmented.
Indexes:
Index name          Owner     Type    Cluster  Columns

pstidx_1            informix  dupls   No       a_sub_num

pstidx_2            informix  dupls   No       a_sub_imsi

pstidx_3            informix  dupls   No       a_sub_imei

pstidx_4            informix  dupls   No       b_sub_num
                                                                b_sub_imsi

b_sub_imei

pstidx_5            informix  dupls   No       ch_start_time

A C-program runs every two hours. At each run about 75000 records are
inserted
into the table. This amount will increase significantly in future.
The C-program gets a file_name passed through by a Unix shell script. It
connects to the database and checks the file name in a table with only 2
fields
(file_name and status). If it finds the file_name the data from that file
is not
been loaded again. If it is a new file the file_name is inserted in the
file_name table. Each run loads data from 32 files.
The data from the flatfile is read into a C-structure. The MAXSIZE of the
structure is set to 10000.
The ESQL/C variables are filled from the structure.
A statement is prepared:
 sprintf(insert_query," INSERT INTO %s
VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

    EXEC SQL PREPARE insert_cdr FROM :insert_query;
    EXEC SQL EXECUTE insert_cdr USING
                                :record_type,
                                :a_sub_num,
                                :a_sub_imsi,
                                ................
When the amount of records > MAXSIZE the program stays in the loop until it
encounters EOF .

Before the indexes were created the program took about 10 minutes to run.
With
the indexes it runs to slow to be acceptable.

Any suggestions are welcome!

Thanks,

Inge Dillen


 
 
 

ESQL/C -Insert- Performance

Post by Daniel Bonesan » Sun, 31 Dec 1899 09:00:00


Have you tried fragmenting the table and using parallel processing ?
Round robin is the best policy for this sort of thing.

>  N. Daniel Bonesana
> SIEMENS-ITRON BUSINESS SERVICES S.A.
> Balcarce 683, (C1064AAM) Buenos Aires, Argentina.

-----Original Message-----

Sent: Friday, March 24, 2000 7:25 AM

Subject: ESQL/C -Insert- Performance

Hi all,

Does anyone know some tricks to improve the performance of an insert into a
big
table.

Here is some info:
We run Informix Dynamic Server Version 7.31.UC3
The table has currently 8214357 rows in it.
Row size=265
The table is fragmented.
Indexes:
Index name          Owner     Type    Cluster  Columns

pstidx_1            informix  dupls   No       a_sub_num

pstidx_2            informix  dupls   No       a_sub_imsi

pstidx_3            informix  dupls   No       a_sub_imei

pstidx_4            informix  dupls   No       b_sub_num
                                                                b_sub_imsi

b_sub_imei

pstidx_5            informix  dupls   No       ch_start_time

A C-program runs every two hours. At each run about 75000 records are
inserted
into the table. This amount will increase significantly in future.
The C-program gets a file_name passed through by a Unix shell script. It
connects to the database and checks the file name in a table with only 2
fields
(file_name and status). If it finds the file_name the data from that file is
not
been loaded again. If it is a new file the file_name is inserted in the
file_name table. Each run loads data from 32 files.
The data from the flatfile is read into a C-structure. The MAXSIZE of the
structure is set to 10000.
The ESQL/C variables are filled from the structure.
A statement is prepared:
 sprintf(insert_query," INSERT INTO %s
VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

    EXEC SQL PREPARE insert_cdr FROM :insert_query;
    EXEC SQL EXECUTE insert_cdr USING
                                :record_type,
                                :a_sub_num,
                                :a_sub_imsi,
                                ................
When the amount of records > MAXSIZE the program stays in the loop until it
encounters EOF .

Before the indexes were created the program took about 10 minutes to run.
With
the indexes it runs to slow to be acceptable.

Any suggestions are welcome!

Thanks,

Inge Dillen


 
 
 

ESQL/C -Insert- Performance

Post by William Ric » Sun, 31 Dec 1899 09:00:00


Just to make sure, how many records were in the table when you
didnt have indexes?  Growth of the file_name table could cause this kind
of performance.  If that is the case, there are paths you might follow
to speed that up.

Will


>Hi all,

>Does anyone know some tricks to improve the performance of an insert into a
big
>table.

>Here is some info:
>We run Informix Dynamic Server Version 7.31.UC3
>The table has currently 8214357 rows in it.
>Row size=265
>The table is fragmented.
>Indexes:
>Index name          Owner     Type    Cluster  Columns

>pstidx_1            informix  dupls   No       a_sub_num

>pstidx_2            informix  dupls   No       a_sub_imsi

>pstidx_3            informix  dupls   No       a_sub_imei

>pstidx_4            informix  dupls   No       b_sub_num
>                                                                b_sub_imsi

>b_sub_imei

>pstidx_5            informix  dupls   No       ch_start_time

>A C-program runs every two hours. At each run about 75000 records are
inserted
>into the table. This amount will increase significantly in future.
>The C-program gets a file_name passed through by a Unix shell script. It
>connects to the database and checks the file name in a table with only 2
fields
>(file_name and status). If it finds the file_name the data from that file is
not
>been loaded again. If it is a new file the file_name is inserted in the
>file_name table. Each run loads data from 32 files.
>The data from the flatfile is read into a C-structure. The MAXSIZE of the
>structure is set to 10000.
>The ESQL/C variables are filled from the structure.
>A statement is prepared:
> sprintf(insert_query," INSERT INTO %s VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?
>,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",table_name);

>    EXEC SQL PREPARE insert_cdr FROM :insert_query;
>    EXEC SQL EXECUTE insert_cdr USING
>                                :record_type,
>                                :a_sub_num,
>                                :a_sub_imsi,
>                                ................
>When the amount of records > MAXSIZE the program stays in the loop until it
>encounters EOF .

>Before the indexes were created the program took about 10 minutes to run.
With
>the indexes it runs to slow to be acceptable.

>Any suggestions are welcome!

>Thanks,

>Inge Dillen


------------------------------------------------------------
This e-mail has been sent to  you courtesy of OperaMail,  as a  free  service  from
Opera  Software,  makers  of the award-winning Web Browser, Opera. Visit us at
http://www.opera.com/ or our portal at: http://www.myopera.com/ Your free e-mail
account is waiting at: http://www.operamail.com/
------------------------------------------------------------
 
 
 

ESQL/C -Insert- Performance

Post by Obnoxio The Clow » Sun, 31 Dec 1899 09:00:00




> > Try dropping some or all of your indexes and then recreate them after
>the
> > inserts have finished.
> > You may find that this is quicker. You may also have to run an update
> > statistics on your table
> > after you've re-created your indexes.

>I am not too clear on this Dude. Are you saying that every 2 hours, they
>should drop all indexes, load 75000 records, then create 5 indexes on over
>8 million records, and all in under 2 hours?

Yep, that's what he said. And of course, render the database inaccessible
during the process.

I'm guessing by the domain (orange.be) that this is a company Rekaish should
have a useful answer for. I'd honestly hope so! :-)

To return to the original question: why worry about batch loading? Why not
trickle feed the table continuously? Give the loading program a low priority
and set lock mode to wait...
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

 
 
 

ESQL/C -Insert- Performance

Post by Rekaish Bhardwa » Sun, 31 Dec 1899 09:00:00


I have very few Answers.
To be honest, I'm just struggling to keep up with you guys (No compliment
intended)
Don't take the proverbial too often.  
Although I can take it, it doesn't do my confidence any good.
I'm doing my best. I usually have a stoke of genius about once or twice a
year.
Unfortunately, this years' stroke, has already been and gone.

orange.be (never heard of them. Sounds like some sort of foreign fruit
producer)

-----Original Message-----

Sent: 24 March 2000 14:13


Subject: RE: ESQL/C -Insert- Performance



> > Try dropping some or all of your indexes and then recreate them after
>the
> > inserts have finished.
> > You may find that this is quicker. You may also have to run an update
> > statistics on your table
> > after you've re-created your indexes.

>I am not too clear on this Dude. Are you saying that every 2 hours, they
>should drop all indexes, load 75000 records, then create 5 indexes on over
>8 million records, and all in under 2 hours?

Yep, that's what he said. And of course, render the database inaccessible
during the process.

I'm guessing by the domain (orange.be) that this is a company Rekaish should

have a useful answer for. I'd honestly hope so! :-)

To return to the original question: why worry about batch loading? Why not
trickle feed the table continuously? Give the loading program a low priority

and set lock mode to wait...
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

 
 
 

ESQL/C -Insert- Performance

Post by Obnoxio The Clow » Sun, 31 Dec 1899 09:00:00



Quote:

>I have very few Answers.

And none on Computers, eh? :-)

Quote:>To be honest, I'm just struggling to keep up with you guys (No compliment
>intended)

None taken.

Quote:>Don't take the proverbial too often.

Awww, shucks. I've hurt his feelings. Sorry, mate.

Quote:>Although I can take it, it doesn't do my confidence any good.
>I'm doing my best. I usually have a stoke of genius about once or twice a
>year.
>Unfortunately, this years' stroke, has already been and gone.

>orange.be (never heard of them. Sounds like some sort of foreign fruit
>producer)

Jeez mate, the future is...? And Computer Answers provides...? And he wants
to load up huge batches of data on a regular basis...? Shall I get out my
crayons now? :-)
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
 
 
 

1. ESQL/C -Insert- Performance

Sorry folks, typo error, Should have been Online 5.

server.  You can play with the number of child's to fork.  This is the best
way to load data fast on a multi CPU server with Online 7.  I am not to
sure if you will get the same improvement with IDS 7x+

2. IDMS 2000: Call for Posters and Demonstrators

3. US- MN Unix Sys Performance Analystl - ESQL/Oracle $70 - $80k, plus benefits

4. Type 4 drivers for MSSQL 7.0

5. Relative performance of OCI versus ESQL/C (PRO*C)

6. printing more than 80-char in a single line

7. ESQL performance

8. complicated SQL - need ideas

9. ESQL Performance & NoLocks

10. ESQL bug & performance isssue

11. ESQL - performance

12. query performance (esql vs dbaccess)