Software Systems
Spring 2005

For today, you should have:

1) worked on your project

2) read Raskin


Outline:

1) file system interface

2) Raskin

3) C / pthreads debriefing

4) project time?


For next time you should:

1) work on your project

2) read Database System Implementation and answer reading questions

3) prepare for a quiz




Reading Questions
-----------------

Database System Implementation, pp 40-62

There are more page in this reading than usual, but at least
on the first pass, you can skip the running Example and just
read the main text.

On a second pass, you might want to read the examples to test
your understanding.

There is a little database-specific vocabulary, so here are
some definitions that might help.

A "relation" is like a dictionary or hash table that maps keys
  to values.  The keys are often unique identifiers of some
  kind, the values are often records.

A "tuple" is a value stored in a relation, also called a record.

If you took Software Systems with me, we saw an example of
a database with two relations, one for respondents and one for
intervals.  Each relation mapped a unique identifier to a
tuple, or record, that stored all the information about a given
respondent or interval.


1) What is the primary performance assumption that underlies
   the design of database algorithms?





2) What work is done during the first phase of a multiway
   merge sort?




3) What work is done during the second phase?




4) During the second phase, why aren't the sublists merged pairwise?




5) Page 47 makes it sound like the two phases take the same amount
   of time, but in practice, which one do you think takes longer?




6) Page 48 makes an argument that two phases is enough to handle the
   biggest conceivable database.  Do you buy it?




The bulleted list on page 51 is an outline of the next few sections.
Come back to this list at the end to make sure you understand each
point.

7) In Section 2.4.2, why is it important that the disks have
   independent heads?




8) In Section 2.4.3, would it make any sense to have two copies
   of the same data on the same disk?




9) What is the primary advantage of the elevator algorithm over
   FIFO?




10) What is the primary drawback of the elevator algorithm?