Information Technology Reference
In-Depth Information
another avenue to be followed so as to complete the transaction. In fact, with partial
rollbacks, every transaction can be programmed to terminate with a
commit
request
and never with a
rollback
request (i.e., total rollback). The effect of a total rollback
can be achieved by setting a savepoint before the first update and then, at the end,
by performing a partial rollback to that savepoint and committing the transaction.
1.10
Multiple Granularity
We may extend our key-range transaction model by adding
multiple granularity
,so
that tuples can be grouped into relations. The relations of the database constitute a
set that is totally ordered according to their identifiers. We denote this set by
f
.r
1
;R
1
/;:::;.r
n
i
;R
n
i
/
g
,
where r
i
is an identifier that uniquely identifies a relation in the database and R
i
is
the relation schema (= X
i
V
i
). Tuple .x;
v
/ in relation r is then uniquely identified
by the pair .r; x/.
The tuple-wise forward-rolling actions in the transaction model are now:
1. RŒr; x;
z
;
v
: reading of the first tuple .x;
v
/ with x
z
from relation r .
2. RŒr; x; >
z
;
v
: reading of tuple .x;
v
/ next to
z
from relation r .
3. WŒr;x;
u
;
v
: update of tuple .x;
u
/ in relation r .
4. IŒr;x;
v
: insertion of tuple .x;
v
/ into relation r .
5. DŒr; x;
v
: deletion of tuple .x;
v
/ from relation r .
New actions include:
(a) RŒr
0
;r;R
0
: browsing the schema R
0
of relation r
0
.
(b) IŒr;R: creation of a new relation r.R/ into the database, corresponding to the
SQL
statement
create table
r.R/.
(c) DŒr; R: deletion of an empty relation r from the database, corresponding to the
SQL
statement
drop table
r for an empty relation r .
Additional levels could be added to the granule hierarchy by grouping relations
into databases (for different owners); new actions would then include ones corre-
sponding to the
SQL
statements
create database
and
destroy database
.
Problems
1.1
The personnel database of an enterprise contains, among others, the relations
created by the following
SQL
statements:
create table
employee
(
empnr
integer not null
,
name
varchar(40) not null
,
address
varchar(80) not null
,
