Archives of the TeradataForum
Message Posted: Tue, 27 Jul 1999 @ 14:56:36 GMT
| Subj: || || Re: Hash buckets |
| From: || || Fred Pluebell |
Row Hash is 32 bits (over 4 billion possible distinct values).
Hash Bucket number is a subset of Row Hash (16 bits in V2), used only to map Row Hash values to AMPs. Yes, there are "only" 64K hash
buckets in the entire system. But that's large enough to be able to distribute rows to AMPs fairly evenly, even on a large multi-node
On the AMPs, the Table ID plus the entire 32-bit Row Hash value are used to locate the correct Cylinder (via memory-resident Master
Index), and then that Cylinder Index is used to locate the Data Block.