Is there any way to improve Query which searches XML data from a table??

Question

hi all,
i have a table which have one column say 'colA' as Varchar(max) datatype which  i  save xml data and it have other cols too 
Currently i am searching data inside this table using like operator
eg:
Select * from tablename where colA like ‘%<tagname>parameterstringvalue</tagname>%’
when i check with the Execution plan i could see it Takes 82% for clusterd index scan ( primarykey col
not ColA)
i added new non clusterd index for the same with include col as ColA and i found nonclusterd index scan
with same estimated I O cost and Extimated operator cost as clusterd index scan
My Question is :-
1. why didnt nonclusterd index seek come?
2. In What way i can improve perfomance for such situvation? i had seen couple of post suggesting to rewrite the Query as SELECT
* FROM myTable WHERE CONTAINS (myCol1, myCol2, "myString").
I Try creating Full text index and found cost increased compared to the original Query ?
3. As per my assumption the wild charecter ('%') in begining makes perfomance issue is there any option
or an alternative for such case?

Answer 1

hi...i can give a skeleton 
--Table Structure------------
Table1:-
(colA - int(PK),
ColB - Varchar(max),
ColC-uniquieidentifier,
ColD-datetime,
ColE-Bit)
It have clusterd index for ColA
Table2:-
(ColA-int(fk)
colF-int(pk)
colG-varchar(max),
ColH-uniqueidentifier,
colI-int,
colJ-int
ColK-date)
-----------------Query Skeleton-------------------: 
select Distinct
s.colA,
s.ColB,
S.colC,
S.colD
from Table1 s with (nolock)
left outer join table2 Q with (nolock) on s.colA=q.ColA
where Q.ColA is null
and s.colB like '%<tag>sometext</tag>%'
and s.colD >='1/1/2010'
and s.colD <='1/1/2014'
i hope this will help to understand it clearly.........