자문 자답 입니다만,
이렇게 해도 될까요?
Update 된 Row 수랑 결과는 맞는거 같은데...

이렇게 해도 될까요? 아니면 다른 방법도 있나요?

------------------------------------------------------------------------------------------------------------------------------------

UPDATE A AA

SET AA.COL3 = (SELECT DISTINCT CASE WHEN B.COL2 IS NULL

                                                                              THEN 'D'

                                                                              ELSE 'C'

                                                                     END COL3

                                      FROM B,

                                                   (SELECT COL1, COL2, COL3 FROM A WHERE 1=1 AND COL3 IS NULL) A

                                     WHERE 1=1

                                            AND A.COL1 = B.COL1(+)

                                            AND A.COL2 = B.COL2(+)

                                            AND A.COL1 = AA.COL1

                                            AND A.COL2 = AA.COL2

                                            AND A.COL3 IS NULL)

WHERE AA.COL3 IS NULL

AND EXISTS (SELECT ' '

                              FROM B,

                                          (SELECT COL1, COL2, COL3 FROM A WHERE 1=1  AND COL3 IS NULL) A

                          WHERE 1=1

                                AND A.COL1 = B.COL1(+)

                                AND A.COL2 = B.COL2(+)

                                AND A.COL1 = AA.COL1

                                AND A.COL1 = AA.COL2

                                AND A.COL3 IS NULL)

;

해당 뷰의 조건항목 col1, col2 에 인덱스가 있는지 모르겠네요.
인덱스를 이용할 수 있어야만 합니다.

UPDATE a
   SET col3 = (SELECT NVL(MIN('C'), 'D')
                 FROM b
                WHERE b.col1 = a.col1
                  AND b.col2 = a.col2
                  AND ROWNUM = 1
               )
 WHERE col3 IS NULL
;

이 스칼라서브쿼리 방식의 경우 인덱스가 없다면 치명적입니다.
  - 150만건 풀스캔을 6600번 반복 수행
이 경우엔 조인 결과를 이용하여 Merge 하는 방법이 좋겠네요.

MERGE INTO a
USING
(
SELECT a.col1, a.col2
     , DECODE(COUNT(b.col1), 0, 'D', 'C') col3
  FROM a
     , b
 WHERE a.col1 = b.col1(+)
   AND a.col2 = b.col2(+)
   AND a.col3 IS NULL
 GROUP BY a.col1, a.col2
) b
ON (a.col1 = b.col1 AND a.col2 = b.col2)
WHEN MATCHED THEN
UPDATE SET col3 = b.col3
;