Consider an extendible hash index with all of directory pages (storing pointers to buckets), bucket pages (storing data entries), and data pages (storing data records) on disk. The index uses Alternative (2) for data entries. Assume there is no overflow page.

There are 1,000,000 records.

Each record has 100 Bytes.

Each data entry has 20 Bytes.

Each directory entry has 10 Bytes.

A disk page has 4000 Bytes.

Each bucket is 1/2 page full.

(1) What is the number of data pages?

(2) What is the number of bucket pages?

(3) What is the number of directory pages?

(4) What is number of I/O pages required to retrieve a record by a search key value k using this index, assuming that the search key is a candidate key?

(5) In (4), what if the search key is not a candidate, that is, how do you determine the I/O cost?