
M =
the number of
memory buffers in
main memory that is
available to
an operator
Note:
 We do not include
the memory space used to
hold the
output of the
operator in the
calculation !!!


B(R) =
the number of
disk blocks
accessed/used for
the relation R
Notes:
 We use B for short 
when it is clear which
relation is referred to.
 B(R) is a
good estimate for
the # disk IO operations
to scan (read) a
cluster relation R


T(R) =
the number of
tuples
accessed/used for
the relation R
Notes:
 We use T for short 
when it is clear which
relation is referred to.
 T(R) is a
good estimate for
the # disk IO operations
to scan (read) a
uncluster relation R


V(R, attr) =
the number of
distinct values found
in attribute "attr" in
relation R
Example:
Employee:
fname lname dno
  
John Smith 5
Frankl Wong 5
Alicia Zelaya 4
Jennif Wallace 4
Ramesh Narayan 5
Joyce English 5
Ahmad Jabbar 4
James Borg 1
V(Emp, dno) = 3 (3 distinct values: 1, 4, 5)

Note:
 V(R, attr)
=
 δ ( π_{attr}(R)
) 


V(R,
[a_{1},a_{2}, ..,a_{n}]) =
the number of
distinct values found
in attributes
"a_{1},a_{2}, .., a_{n}" in
relation R
Note:
 V(R,
[a_{1},a_{2}, .., a_{n}] )
=
 δ ( π_{
a1,a2, .., an
}(R) ) 

