Processor hungry functions

Question

Hello
I have a table of 1.3 m records that contains about 800k unique companies.
We use this table to process all our in house data for matching, it has various functionsrun on certain columns to prepare it so we can match incoming data.
Recently the functions I have in place have become very processor hungry, very slow and are creating way too many redo log and I desperatley need some help on how to improve these as it is becoming a core part of our business.
here is an example, this function standardises all Address data by comparing a column to a look up table, it takes far too long. the lookup table ahs about 60k records in it.
This is just one of about 12 functions, I have more based on this example which create algorithims on company names for matching
Im new to all this so if you need this in a different format please let me know
thanks for your help in advance
h1. code
update statement
update boss_universe set st_addr1 = site_basura(st_addr1) where st_addr1 is not null;
function called by the update statement
FUNCTION SITE_BASURA
P_STRING IN VARCHAR2 DEFAULT null
) RETURN VARCHAR2 AS
type ent_srch is table of b_entity.srch_word%type;
type rpcWth is table of b_entity.replacewith%type
index by pls_integer;
l_cnt ent_srch;
l_rpcWth rpcWth;
entity varchar2 (1000);
strngFound varchar2(1000);
lroot varchar2(1000);
lentity varchar2(1000);
sqlstring varchar2(1000);
sqlstring2 varchar2(1000);
v_entity b_entity%rowtype;
t_string varchar2(1000);
p_root varchar2(1000);
l_cnt ent_srch;
l_rpcWth rpcWth;
v_entity b_entity%rowtype;
t_string varchar2(1000);
BEGIN
if P_STRING is not null then
lroot:=lower(replace(P_STRING,chr(39)));
p_root:=lroot;
for x in ( select
lower(srch_word) srchPrbsra,
replacewith as rplcPrbsra
from
PREBASURA
where
category ='SITE' and
' '||p_root||' ' like '% '||lower(srch_word)||' %')
loop
p_root:=replace(' '||p_root||' ',' '||x.srchPrbsra||' ',' '||x.rplcPrbsra||' ');
end loop;
for x in ( select
lower(srch_word) srchBsra,
replacewith as rplcBsra
from
BASURA
where
p_root like '%'||lower(srch_word)||'%')
loop
p_root:=replace(p_root,x.srchBsra,x.rplcBsra);
end loop;
for x in ( select
lower(srch_word) srchPrbsra,
replacewith as rplcPrbsra
from
PREBASURA
where
category = 'SITE' and
p_root like '%'||lower(srch_word)||'%')
loop
p_root:=replace(p_root,lower(x.srchPrbsra),x.rplcPrbsra);
end loop;
for x in ( select
lower(srch_word) srchPrbsra,
replacewith as rplcPrbsra
from
PREBASURA
where
category ='SITE' and
p_root like '%'||lower(srch_word)||'%')
loop
p_root:=replace(p_root,x.srchPrbsra,x.rplcPrbsra);
end loop;
for x in ( select
lower(srch_word) srchPrbsra,
replacewith as rplcPrbsra
from
PREBASURA
where
category is null and
p_root like '%'||lower(srch_word)||'%')
loop
p_root:=replace(p_root,x.srchPrbsra,x.rplcPrbsra);
end loop;
for x in ( select
lower(srch_word) srchPrbsra,
replacewith as rplcPrbsra
from
PREBASURA
where
category ='SITE' and
' '||p_root||' ' like '% '||lower(srch_word)||' %')
loop
p_root:=replace(' '||p_root||' ',' '||x.srchPrbsra||' ',' '||x.rplcPrbsra||' ');
end loop;
end if;
RETURN p_root;

Answer 1

Hi guys thanks for your replys, Ill try the CACHE route, problem is we have to do this with the address data as it is creating algoritihims for matching
Im self taught so bear with me.
Bascially we have in house data of around 5m records, we are trying to create a tool that standardises all that data ready for mathing, so the address columns for example all need to be one standard. E.g Street becomes Str, House becomes Hse, Avenue becomes Ave and so on. We do the same with company names when E.g Limted becomes Ltd Ltd becomes Ltd, this way we can standardise all the data.
We do this using lookup tables as there are thousands of variations, my idea was instead of writing all those variations into code, write one line of code (the function) and have it loop thorugh those look up tables. Problem is its so slow and I dont know how to speed it up
any more ideas wouldbe amazing
thanks for your help