---

Intro to indexing using hashing

---

---

---

 

Slideshow:

Review: Hashing

A good hash function will distribute keys uniformly over the range [0..(B−1)]
Review: Hashing as a map (memory) data structure

Hash index: an index file organized using hashing (memory) data structure

Hash index: implementation

Hash index: example

Sample use: find record (ptr) for give search key

Find the record ptr for search key W
Sample use: find record (ptr) for give search key

Hash the search key W -- suppose Hash(W) = 1
Sample use: find record (ptr) for give search key

Read the index block (and possible overflow blocks) and search for key W
Sample use: find record (ptr) for give search key

If found, use the record ptr to access the data record - otherwise, data is not in file
Inserting a new search key (+ rec ptr) in a hash index file

Deleting a search key (+ rec ptr) from a hash index file

❮ ❯



---

---

---

---

• Hash Function
  • Hash Function:

    Hash function = a function that maps a search key to an index between [0 .. B-1] (B = the size of the hash table)

    ---

  • Bucket:

    Bucket = a location (slot) in the bucket array

    ---

  • The hashing technique:

    Note:

    Different search keys can be hashed into the same hash bucket !!!

  ---

  ---

  ---

  ---

• Hashing used as an indexing technique
  • How to use use Hashing as a indexing technique to find records stored on disk:

    Bucket Array contains a (search key, block/record pointer) map object in each bucket:

    ---

    ---

  • The Bucket Array:

    The bucket array is stored on disk: 1 bucket = 1 disk block 1 bucket will contain n (search key, block ptr) !!!

  ---

  ---

  ---

  ---

• Storing hash tables on disk
  • Recall that a file entry in a directory contains a list of block numbers:

    B block numbers <-------------------> +-----------+-------------------+-+-+-+-+-+ ...... +-+ | Filename | owner/access info | | | | | | | | +-----------+-------------------+-+-+-+-+-+ ...... +-+

    ---

    ---

  • Each data block of a Hash index file is one bucket:

    The Hash index file has B buckets

  ---

  ---

  ---

  ---

• How to find data records using a Hash Index
  • Sample hash index file:

    How to use a Hash index to find the data record with search key W:

    Find the data record containing the search key W: Steps: Compute the hash value h(W) Read the disk block(s) of bucket h(W) into memory Search (a linear search algorithm is sufficient because memory search is fast) the bucket h(W) for: W, RecordPtr(W) Use RecordPtr(W) to access W on disk

  ---

  ---

  ---

  ---

• Inserting into a Hash index
  • Insert Algorithm:

    Insert( K, recordPtr(K) ) { k = h(K); // Hash the key Find the hash bucket (= a CHAIN of disk blocks !!!) if ( hash bucket has space ) { insert (K, recordPtr(K)) in bucket; } else { allocate an overflow block; insert (K, recordPtr(K)) in overflow block; } }

  ---

  ---

  ---

  ---

• Deleting from a Hash table
  • Delete Algorithm:

    Delete( K ) { k = h(K); // Hash the key Find the hash bucket (a CHAIN of disk blocks !!!) if ( x is found in hash bucket ) { Delete (K, recordPtr(K)) from bucket; } Optional: Consolidate overflow blocks of a bucket into few blocks }

    Delete oscillation caveat:

    Consolidation must be done with care because: Inserts after deletes can cause oscillation !!! Oscillation: Insert: a block splits Delete: 2 blocks merges Insert: a block splits again Delete: 2 blocks merges again... .... ( Oscillation --- can result in poor performance !!!)