Sqlloader : is there any performance issue ?

Question

hI fRIENDS,
The below is sqlldr command ( i am using direct path ) and control file .
sqlldr
sqlldr $databaseConnectionString parfile=$parameterFile log=$logPath/sqlloader.log bad=$logPath/sqlloader.bad direct=true
control file (long one :))
LOAD DATA
INFILE 'fact.dat' "str '\n'"
INTO TABLE sqlldr_feed_fact_mtd
TRUNCATE
WHEN VW = 'Periodic'
FIELDS TERMINATED BY ';'
(PERIOD POSITION(1) CHAR ,
VW CHAR,
AMOUNT CHAR,
BRANCH CHAR,
ACCOUNT CHAR,
VALUE1 CHAR,
VALUE2 CHAR,
VALUE3 CHAR,
CUSTOM4 CHAR,
PARTY CHAR,
DDATA INTEGER EXTERNAL)
INTO TABLE sqlldr_feed_fact_qtd
TRUNCATE
WHEN VW = 'QTD'
FIELDS TERMINATED BY ';'
(PERIOD POSITION(1) CHAR,
VW CHAR,
AMOUNT CHAR,
BRANCH CHAR,
ACCOUNT CHAR,
VALUE1 CHAR,
VALUE2 CHAR,
VALUE3 CHAR,
CUSTOM4 CHAR,
PARTY CHAR,
DDATA INTEGER EXTERNAL)
INTO TABLE sqlldr_feed_fact_ytd
TRUNCATE
WHEN VW = 'YTD'
FIELDS TERMINATED BY ';'
(PERIOD POSITION(1) CHAR,
VW CHAR,
AMOUNT CHAR,
BRANCH CHAR,
ACCOUNT CHAR,
VALUE1 CHAR,
VALUE2 CHAR,
VALUE3 CHAR,
CUSTOM4 CHAR,
PARTY CHAR,
DDATA INTEGER EXTERNAL)
number of rows in date file is 847055. there can be more rows in near future for that .dat file.
when i execute above - it takes 12mins... is it ok ??
please give ur comment!

Answer 1

I can load 3 million rows with direct path load in well under 1 minute.
I guess there are several factors involved in performance where SQL*Loader is concerned.
e.g.
1. is the data residing on the server and SQL*Loader being run from the server? If so, it should be fast. If the data however resides on a client machine and SQL*Loader is being run from the client then you will have network delay for all the data being sent over the network. Even worse if the data resides on a networked drive and SQL*Loader is being run from a client as the data has to be fetched over the network, to be sent back over the network to the database server.
2. What is the hardware spec of the server? Obviously the more powerful the server, the faster disks it has etc. the faster the data can be loaded.
3. You're loading data conditionally to split it into different tables. That will add a slight overhead to the processing.
etc.