Slideshow:
Solving the block identification/location:
Translation Table
Solving the block identification/location:
Translation Table
Structure/use of the Translation table:
Example of using the Translation table
How to
update
the Translation table
How to
update
the Translation table
How to
use
the Translation table
Problem: we want to access a block/record identified by its database address x
How to
use
the Translation table
Problem: we want to access a block/record identified by its database address x
How to
use
the Translation table
Problem: we want to access a block/record identified by its database address x
How to
use
the Translation table
Problem: we want to access a block/record identified by its database address x
Example 1 on
using
the Translation table
Example 1 on
using
the Translation table
Example 2 on
using
the Translation table
Example 2 on
using
the Translation table
- Translation Table:
- Translation table = a
(hashing-based) data structure used to
solve the
following 2 problems:
- Check if
a disk block/record is
stored
in memory
- If a
disk block/record is
stored
in memory,
the translation table can
be used to:
- find the memory address of the disk block/record (in memory) using the database address
by defining the following mapping function:
Database Address ----> Virtual Memory Address
- Check if
a disk block/record is
stored
in memory
- Translation table = a
(hashing-based) data structure used to
solve the
following 2 problems:
- Initial situation:
- The (disk) record is
only stored
on disk
- The (disk) record is
only stored
on disk
- What happens when you
read
a disk block into
computer memory:
- Suppose we
read the
disk block "Block 2" into
the memory
Let: xxxx = the (virtual) memory address of the memory location of the block in memory
- Then: store the
(DB-addr, Virtual-addr)
into the (hash) translation table:
- Suppose we
read the
disk block "Block 2" into
the memory
- How to
use the
translation table
to access a
database object
with its database address
x:
- Hash the
database address
x
- If database address
x is
found in
translation table:
- A copy of
the data item (disk block or record) is
stored
in memory !!!
- We must use the corresponding virtual memory address to access database object in memory
- A copy of
the data item (disk block or record) is
stored
in memory !!!
- If database address
x
is
not found in
translation table:
- The data item (block or record) is
not found
in memory !!!
Therefore:
- You must first read the
database object
(from disk) into memory
(The read operation must update (= maintain) the translation table with the virtual memory address of the database object in memory)
- You must first read the
database object
(from disk) into memory
- The data item (block or record) is
not found
in memory !!!
- Hash the
database address
x
- Example 1:
accessing a record
that is
already stored
in memory:
- Suppose we access
the record (with address "xyz")
and
the record
is in memory:
- We hash "xyz"
and will find
a valid entry in the
translation table:
- We will use the corresponding memory address to access the copy in memory
- Suppose we access
the record (with address "xyz")
and
the record
is in memory:
- Example 2:
access a record
that is not
presently in memory:
- Suppose we access
the record (with address "xyz")
and
the record
is not
stored in memory:
- We hash "xyz"
and will find
a invalid entry in the
translation table:
- This will trigger a
read operation
to bring the
database object into
memory
(And updates the translation table !!!)
(Note: we know the database address - so we can find the block on disk)
- This will trigger a
read operation
to bring the
database object into
memory
- Suppose we access
the record (with address "xyz")
and
the record
is not
stored in memory:
- Why use a
hash table:
- The translation Table is
organized as
a Hash Table because:
- Relatively few
disk blocks will be
found in
memory
(Disks are T bytes in size. The memory is only G bytes in size. So only a small fraction of disk blocks is in memory)
- The translation table
only need to
store
information on
disk blocks that are
in memory
(So the hash table size can be relatively small)
- The translation table
only need to
store
information on
disk blocks that are
in memory
- A small number of
hash entries will
result in a sparse hash table
Hash table is very efficient if the table is sparse (because almost no collision) !!!
Graphically:
- Relatively few
disk blocks will be
found in
memory
- The translation Table is
organized as
a Hash Table because: