How to combine update stmts with different filters and multi-columns?

Question

Hi,
I found these following update stmts in one of our long running jobs. I'd like to see if it is possible to combine it in one update or merge stmt so I can reduce the scan time on the same table from 4 times to 1 time. But I'd need some expert suggestions for how to do that. Here are the stmts:
UPDATE table1 c
SET (polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,valid)=(SELECT DISTINCT polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,'A'
FROM table2 v
WHERE v.polnum=c.pol# AND
v.polren=c.ren# AND
v.polseq=c.seq# AND
v.vehnum=c.veh#)
WHERE c.polnum IS NULL;
UPDATE table1 c
SET (polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,valid)=(SELECT DISTINCT polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,'B'
FROM table2 v
WHERE v.polnum=c.pol# AND
v.polren=c.ren# AND
v.polseq=c.seq# AND
v.mainveh='Y')
WHERE c.polnum IS NULL;
UPDATE table1 c
SET (polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,valid)=(SELECT DISTINCT polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,'C'
FROM table2 v
WHERE v.polnum=c.pol# AND
v.polren=c.ren# AND
v.polseq=0 AND
v.mainveh='Y')
WHERE c.polnum IS NULL;
UPDATE table1 c
SET (polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,valid)=(SELECT DISTINCT polnum,polren,polseq,vehnum,
compgrp,prodgrp,uwsys,comp,
state,prod,rrdate,brand,
agycode,AGYCLASS,'D'
FROM table2 v
WHERE v.polnum=c.pol# AND
v.polren=0 AND
v.polseq=0 AND
v.mainveh='Y')
WHERE c.polnum IS NULL;
Table1 has 800000 rows, Table2 has 200 million rows. In table1, about 12.5% rows has a null polnum, so 12.5% data will be updated.
Thank you in advance with all your suggestions!

Answer 1

the reason I never specified the joins is because I thought the joins were there to derive the Valid indicator, in which case they are not required if you are using the CASE statement. The join, "V.Polnum = C.Pol#" is still required as this is across all rows! If that is the case, then it can be simplified even more so;
update   Table1 C
set      (Polnum,Polren,Polseq,Vehnum,Compgrp,Prodgrp,Uwsys,Comp,State,Prod,Rrdate,Brand,Agycode,Agyclass,Valid) =
            ( select   distinct Polnum
                               ,Polren
                               ,Polseq
                               ,Vehnum
                               ,Compgrp
                               ,Prodgrp
                               ,Uwsys
                               ,Comp
                               ,State
                               ,Prod
                               ,Rrdate
                               ,Brand
                               ,Agycode
                               ,Agyclass
                               ,case
                                  when (Polren + Polseq=0) and (Mainveh='Y') then
                                        'D'
                                  when (Polren=0) and (Polseq>0) and (Mainveh='Y') then
                                        'C'
                                  when (Polren>0) and (Polseq>0) and (Mainveh='Y') then
                                        'B'
                                  when (Polren>0) and (Polseq>0) and (Mainveh!='Y') then
                                        'A'
                                end as Valid
             from      Table2 V
             where     V.Polnum = C.Pol#
where    C.Polnum is null;However if you are returning more rows than what you are expecting from the above update statement because the joins are being used to limit the data as well, then this might be more appropriate:
update   Table1 C
set      (Polnum,Polren,Polseq,Vehnum,Compgrp,Prodgrp,Uwsys,Comp,State,Prod,Rrdate,Brand,Agycode,Agyclass,Valid) =
         ( select v2.PolNum
               ,v2.Polren
               ,v2.Polseq
               ,v2.Vehnum
               ,v2.Compgrp
               ,v2.Prodgrp
               ,v2.Uwsys
               ,v2.Comp
               ,v2.State
               ,v2.Prod
               ,v2.Rrdate
               ,v2.Brand
               ,v2.Agycode
               ,v2.Agyclass
               ,case
                  when (v2.Polren + v2.Polseq=0) and (v2.Mainveh='Y') then
                        'D'
                  when (v2.Polren=0) and (v2.Polseq>0) and (v2.Mainveh='Y') then
                        'C'
                  when (v2.Polren>0) and (v2.Polseq>0) and (v2.Mainveh='Y') then
                        'B'
                  when (v2.Polren>0) and (v2.Polseq>0) and (v2.Mainveh!='Y') then
                        'A'
                end as Valid_Flag
            ( select   distinct Polnum
                               ,Polren
                               ,Polseq
                               ,Vehnum
                               ,Compgrp
                               ,Prodgrp
                               ,Uwsys
                               ,Comp
                               ,State
                               ,Prod
                               ,Rrdate
                               ,Brand
                               ,Agycode
                               ,Agyclass
                               ,Mainveh
             from      Table2 V
             where     V.Polnum = C.Pol#
             ) v2
         where (v2.Polren = C.Pol# or v2.Polren = 0)
         and   (v2.Polseq = C.Seq# or v2.Polseq = 0)
where    C.Polnum is null;As before, both remain UNTESTED!