| Oracle Single Table Hash Clusters (merged) [message #388355] |
Tue, 24 February 2009 07:54  |
pramodkp
Messages: 4
Registered: February 2009
|
Junior Member |
|
|
How does Oracle determine the block address by applying hash on cluster key? Is some meta data [something like a hash table] maintained in the cluster for this purpose ?
If yes,
what is the overhead per hash key?
is there a way to pin it in the buffer cache?
|
|
|
|
|
|
| Re: Oracle Single Table Hash Clusters (merged) [message #388453 is a reply to message #388357] |
Tue, 24 February 2009 22:13  |
pramodkp
Messages: 4
Registered: February 2009
|
Junior Member |
|
|
Thanks for the pointers, Mike.
But I don't find the answers there. It states that "...resulting hash value corresponds to a data block in the cluster, which Oracle then reads/writes on behalf of the issued statement"
I am keen to know if hash-tables are maintained somewhere and the total overhead per key.
Regards,
Pramod K P
|
|
|
|
|
|
|
|
|
|
|
|
| Re: Oracle Single Table Hash Clusters (merged) [message #388556 is a reply to message #388546] |
Wed, 25 February 2009 06:52 |
 |
Michel Cadot
Messages: 68728
Registered: March 2007
Location: Saint-Maur, France, https...
|
Senior Member
Account Moderator |
|
|
Oh yes, my apologize I mixed up B*Tree and Hash clusters.
You are right the hashing function does determine the block.
You don't need any hash table, the hash function "IS" the logical hash table and unless you have too much collisions you only need 1 LIO.
Regards
Michel
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
