Presentation is loading. Please wait.

Presentation is loading. Please wait.

RUM Conjecture of Database Access Method

Similar presentations


Presentation on theme: "RUM Conjecture of Database Access Method"— Presentation transcript:

1 RUM Conjecture of Database Access Method

2 Database Access Method
Defines how to store and access data in a database system Heap file Sorted file B+ Tree index Hash Index

3 Requirements of A Good Access Method
Read optimized how much time it cost to read the data e.g., range search, equality search Update optimized how much time it cost to update the data e.g., insert, delete, update Memory (Space) optimized how much space it uses to store the data e.g., auxiliary structures, duplicates

4 However… RUM Conjecture
RUM Conjecture: An access method that can minimize two out of the read, update, and memory overheads, cannot optimize the third overhead! In other words, there might not be the “best” access method that is optimal for all the three requirements!

5 Example: Append-Only Log
Read unoptimized A search may need to scan the entire log Space unoptimized When delete/update an entry, old entry is not deleted in-place Update optimized Each update simply append the entry to the log

6 Example: B+Tree Read optimized: Update unoptimized: Space unoptimized:
Each search only visits logB(N) pages Update unoptimized: An update requires visiting at least logB(N) pages Space unoptimized: Pages often are not full

7 Example: Hash Index Read optimized: Update unoptimized:
A query simply checks O(1) bucket Update unoptimized: An update may need to split overflow buckets Space unoptimized: Buckets often are not full

8 Other Access Methods


Download ppt "RUM Conjecture of Database Access Method"

Similar presentations


Ads by Google